Ranks not properly updating when overriding a lazer score on stable

[peppy]

Discussed in ppy/osu#29150

^{Originally posted by mooshiros July 28, 2024}
On Cold Weather [Hard] two days ago I set a gold S rank on the daily challenge (3.4 million lazer classic score). I proceeded to override it with a HDDT SS on stable (5 million scorev1, 3.8 million lazer classic score). Despite this, my profile still displays a gold S rank that it shouldn't and one less silver SS than it should. I then proceeded to go back to lazer and set a HD S rank off the daily challenge (3.5 million lazer classic score) and a NM SS on the daily challenge (3.4 million lazer classic score), and yet my profile still displays a gold S rank and one less silver SS than it should. Is there a way to fix this?

Edit 2024-07-31: I confirmed that this is a general issue with lazer rather than just the daily challenge (probably should have done this earlier) by setting a nomod fc on Harumachi Clover [Nino's Insane] on lazer then setting a better hd fc on stable. My profile still displays an extra gold S that it shouldn't, and my silver S count did not increment. Feel like I probably should have made this an issue rather than a discussion but oh well.

[peppy]

Needs further investigation.

[bdach]

I'm not sure there is anything to investigate, the issue seems rather obvious. osu-queue-score-statistics and web-10 live in parallel realities, each sees only their part of scores. osu-queue-score-statistics's rank count processor does not run for legacy scores, and web-10's rank count logic will omit lazer scores because it looks at old tables.

I'm not sure what fixing would entail because that depends on the desired outcome. If rank counts should be uniform regardless of lazer mode then web-10's processing should probably be turned off and queue-score-statistics should take over. If rank counts should change depending on lazer mode then we need new columns or tables to store non-lazer and lazer rank counts separately.

[peppy]

I think we'll want to look at this after we move pp across (shutting down more processing pieces of web-10).