A level 5 rogue will quite probably have a thievery dc of 13, if they invest in it and max dex. The average lock has a dc of 25 and requires 4 successes. It takes a roll of 12 or better to have a single success, and will average about 9 rolls to rack up those 4 successes. With 9 rolls wherein you crit fail on a 2 or lower, the likelihood of breaking a pick is ~61%.
Should a level 5 rogue take a minute to open the average lock, and more likely than not break a pick in the process?
And let's look at a good Lock: DC 30, requiring 5 successes. The level 5 rogue will only succeed on a 17, meaning it will take on average 20 attempts to get those 5 successes. On one attempt in a thousand our Lvl 5 rogue will open this lock before breaking a pick, and will typically break 3 in the process.
I think taking a minute to pick an average lock, or 2 minutes to pick a good lock is pretty reasonable, personally. What's weird is the fact that in PF1 / D&D, it can be done in 6 seconds, most of the time. That's just inhumanly fast. If you break a pick (at a cost of 3sp, assuming common tools and that you just replace the pick rather than bothering to try to repair it), that's not really a big deal - any rogue worth their salt should be carrying spare picks, anyway.
Is it realistic to expect to break a pick on 60% of lock you pick? No, not really, but this is a game, and we already do a lot of things that are unrealistic. As a GM, if I had a player complaining about this, I'd consider letting them take longer to pick the lock (perhaps doubling or tripling the completion time) in exchange for the threshold to critically fail being slightly lower (by 5 points or so), to represent more meticulous work in an effort to specifically avoid breaking tools - but really, the cost is so minimal, I don't really see any reason that this should be needed (except in a case where picks are for whatever reason not readily available, and where time is no real object, such as picking the lock on shackles or something.)
It's a pretty realistic expectation for a mediocre locksmith in the real world faced with the average door lock. It's a bit slow for the fantasy expert lock picking thief who's invested their ability and skill increases to excel at a mundane and achievable task. But time spent is the smaller issue.
And it's 3 gp a thieves tool set, but the bigger issue is bulk. God forbid you've a dozen doors with good locks in a dungeon, that's 4 bulk worth of picks to get through--pretty much the thiefs whole inventory--and a 50 percent chance of ultimate failure (not to mention 240d20). Pretty rough on the class fantasy. If nothing else I'd change the names of the locks to pad the thief ego: poor becomes average, average good, good master and master legendary. I don't want my player stymied by an average door because he only brought one backup toolkit.
Merwyn already pointed out that replacement picks are cheap and weightless. Regarding the number of dice, every PF2 campaign I've been a part of (which is two, so a pretty small sample size) has just had the rogue roll for each tumbler right up front. Need 5 successes, roll 5 dice. If any of them are a critical success, it offsets a critical failure (in addition to counting for two tumblers as normal); otherwise, each critical failure = -1 replacement pick. It's meant on average 2-3 sets of rolls per lock, rather than 7-8, and keeps things moving at a fine clip.