feat: improve compilation speed of CreateOp.lean

[math-fehr]

This was the slowest file to compile so far. I replaced all grind tactics with manual tactics, using simp_getset to make this more readable.

WIP as I want to split this before merging.

[math-fehr]

!radar

[ineol]

Benchmark results for 260501c against 67a1854 are in. There are significant results. @math-fehr

Large changes (13✅, 2🟥)

• ✅ build//instructions: -355.3G (-7.45%)
• ✅ build//task-clock: -47s (-6.45%)
• ✅ build//wall-clock: -29s (-17.70%)
• ✅ build/lakeprof/longest build path//wall-clock: -27s (-16.99%)
• ✅ build/lakeprof/longest rebuild path//wall-clock: -22s (-21.80%)
• 🟥 build/module/Veir.Analysis//instructions: +28.9M (+1.87%)
• ✅ build/module/Veir.Rewriter.GetSet.CreateOp//instructions: -355.8G (-82.42%)
• 🟥 build/module/Veir.Rewriter.WfRewriter.GetSetTactic//instructions: +211.0M (+5.63%)
• ✅ build/profile/elaboration//wall-clock: -4s (-18.69%)
• ✅ build/profile/grind ematch//wall-clock: -12s (-23.58%)
• ✅ build/profile/grind mark subsingleton//wall-clock: -3s (-16.42%)
• ✅ build/profile/grind simp//wall-clock: -23s (-19.20%)
• ✅ build/profile/grind//wall-clock: -34s (-18.52%)
• ✅ radar/run/main//time: -29s (-17.36%)
• ✅ radar/run/main/script//time: -29s (-17.61%)

Medium changes (3✅)

• ✅ build/profile/sym canon//wall-clock: -4s (-21.68%)
• ✅ build/profile/sym typeclass inference//wall-clock: -5s (-12.51%)
• ✅ build/profile/typeclass inference//wall-clock: -3s (-14.81%)

Small changes (6✅, 2🟥)

• 🟥 build/module/Veir.Interpreter.Refinement//instructions: +10.2M (+0.55%)
• ✅ build/module/Veir.Rewriter.GetSet.DetachOp//instructions: -106.9M (-0.33%)
• 🟥 build/module/Veir.Rewriter.WfRewriter.GetSet//instructions: +214.7M (+0.43%)
• ✅ build/profile/.olean serialization//wall-clock: -953ms (-5.71%)
• ✅ build/profile/attribute application//wall-clock: -526ms (-17.43%)
• ✅ build/profile/grind cutsat//wall-clock: -82ms (-20.53%)
• ✅ build/profile/grind ring//wall-clock: -99ms (-14.65%)
• ✅ build/profile/type checking//wall-clock: -3s (-6.70%)

[math-fehr]

@tobiasgrosser @ineol, could one of you review this?