I would go with this reading as well, @Hekazu. For a number of reasons, but my logic is based on a process of procedures in particular. I would and do understand it that you roll to take an action, such as the Attack action, check for advantage (or disadvantage) and take that better (or worse) result, apply other automatic relevant dice such as Bless, have the option to employ any special modifiers such as Bardic Inspiration or Battle Master dice, then check for other miscellaneous dice, then add static bonuses such as one's ability modifier and their proficiency bonus.

Now, a number of these might be in slightly different orders, but I am fairly certain the first two - checking for advantage or disadvantage and then using that singular higher or lower result with whatever rolled automatic bonus after (only once) - are spot on.

You would be right that it changes little, but not being a mathematician I am not entirely sure if one way is better than the other. I am going to err on the side that applying Bless or a similar bonus to the end makes it stronger rather than weaker when advantage is taken into account, but as we've seen that seems to be the developer's intent. For some added intent, we will extrapolate that Bless is an effective +2.5 to attack rolls and saving throws. Given this is Dungeons and Dragons and we round down rather than up, this falls a lot more reasonably in line with a couple of other spells and effects.

Speaking from a paladin's hand at this, which a cleric could see as well, Shield of Faith is a +2 bonus to Armor Class; something that on average, assuming all is equal, not totally mitigate Bless' bonus to attacks. Why might I say that? Well, as the free SRD states:

If the total of the roll plus modifiers equals or exceeds the target’s Armor Class (AC), the attack hits.

So it is obvious they intend for attacks to land more rather than less, otherwise combats become more fruitless flailing about with less hitting for the heroes and the villains alike; this was true in previous editions, but many players houseruled it to "Defenders win on Ties" because it favored the players more. More conjecture on this comes from the fact that Armor Class more or less does not scale but proficiency bonus does. A solid Armor Class of 18 at 1st level means a lot less at say... 20th, where from proficiency alone the enemy only normally needs a 12 now to hit you - assuming they have no other bonuses like from Strength - rather than the previous 16 they needed.

Why is Bless so good?

The real answer for this is because it is a force multiplier. Three players will get more mileage out of it than its counterpart of Bane, because the spell will last longer, up to its concentration duration at most. Targets struck by Bane are most likely going to die sooner, because of being focus fired, and thus the spell's effect "ends" little by little sooner; monsters also tend to be more easily overcome by players in terms of volume, such as flanking against larger units or smaller units unable to survive a sustained assault. A competing spell of interest to a cleric is Hold Person, because if it lands the enemy is almost assured death by the party because all attacks in 5ft are critical hits and made with advantage.

But why isn't Hold Person as good, except against say, a singular enemy and even then? Because that is only one enemy that cleric removes from the fight, assuming they even fail the saving throw. "Solo" encounters, to borrow a Fourth Edition term are unlikely to fail their save and might break free before the party is in position to land their critical strikes. Next is going back to our force multiplier; Bless only gets better the longer combat goes on and the more attacks and saves it affects. Hold Person is also a 2nd level spell, meaning it isn't cheap; they realized how dangerous "Save or Lose" spells were one-and-one-half editions ago.

Bless also likes to favor combinations that are ideal with optimization, such as Great Weapon Master and Sharpshooter, because it offsets their penalty and even more so with advantage, which is more or less the merit we measure them against. Compare this to say, Divine Favor, which adds a +1d4 to all damage rolls. In the end because of how Armor Class functions, hitting more consistently - up to, I think if I recollect correctly ~75% the time or so - is usually better than doing more damage albeit inconsistently. You can check this by looking at the hefty penalties of the two aforementioned feats; its obvious they wanted a risk versus reward element, but went a bit too far with the reward.

But when does Bless taper off?

For your cleric, it loses its luster as the "go-to" concentration spell for combat at 3rd spell level because of Dispel Magic, something most every enemy spellcaster would have, enters the picture. Beacon of Hope also appears here albeit as a more situational spell. But... you can go back an entire spell level and even look at Silence which can eliminate the majority of spells that can even be cast which are the real threat. For paladins? Heroism is a spell you want to cast if you can, because absorbing and mitigating damage is part of your objective as a class. If damage is your objective over a long period, Divine Favor might see you better, but in the end a spell based smite and the smite ability are still stronger because of their capacity to rapidly eliminate single targets. More or less, this means it is essential for paladins to use the Bless spell until you need Magic Weapon and its greater relative, Elemental Weapon.

There's one last major area to cover for Bless and that has to deal with concentration. The majority of clerics have been and are melee centric clerics or at least places where they can be struck. Intelligent enemies who are aware of the cleric or another spellcaster, tend to know those who cast spells are more dangerous than those that do not. But at its core, like anyone concentrating on a spell, the attacked must make a Constitution saving throw at DC 10 or half that of the damage, whichever is higher. Barring the paladin and a Unearthed Arcana sorcerer, no one capable of casting Bless is out of the box going to have Constitution saving throw proficiency; it is going to cost you a feat. Feats are expensive, taking away static bonuses, and not even always an option at some tables.

On a note that's mostly irrelevant for us given the build of our party, Bless favors combat classes. The more offensive oriented players in game that use weapons for attacks, the better the result it has in its returns. A single fighter, a wizard, a cleric and a sorcerer are not going to get much benefit from bless most of the time in comparison.