GS/TC: Optimize block offset calculations.

[lightningterror]

Description of Changes

GS/TC: Optimize block offset calculation.
Swap offset x and y loops:
Having the y loop first allows for better optimizations/caching since it's bigger. Also optimize start point loop conditions for target rect.

Rationale behind Changes

Speed, optimizations.
Fixes #13338

Suggested Testing Steps

Test #13338
Test GTA Liberty City stories, dump on it goes from 43 fps to 135 fps.
Test other games listed in #12794

Did you use AI to help find, test, or implement this issue or feature?

No.

[JordanTheToaster]

Appears to work correctly and brings GTA LCS FPS back up above what it was before the regression.

[lightningterror]

Posting Jordan's results, will be useful for progress report:

Call of duty 2 180 fps to 194 fps
Final Fantasy X 720 fps to 830 fps
GH Van Halen at native 188 fps before 201 fps after
Gun 332 fps before 456 fps after
dot hack infection part 1 619 fps before 912 fps after
Haven call of the king 620 fps to 660 fps
Kung fu panda 415 to 436
LA rush with tex in rt 174 to 180
Midnight club 3 dub remix 290 before 402 after
MLB 11 The Show 639 before 858 after
Need for speed carbon 138 fps before 144 fps after
Project Snowbling 578 fps before 680 fps after
Sakura Taisen - Atsuki Chishioni 983 fps before 1107 fps after
SOTC 187 fps before 206 fps after
Test drive eve of destruction 331 fps before 378 fps after
The god farther 770 fps before 875 fps after

Maybe someone can make some graphs which would be nice.