Migrate/jit/matmul tiling 2d

[louisfd]

Matmul tiling 2d kernels for #1422

Bonus stuff:

• macro for configurable unrolling of loops
• tiling 2d (unpadded) will automatically use the kernel with no bound checks when shapes are divisible by block sizes

[codecov]

Codecov Report

Attention: Patch coverage is 93.61702% with 63 lines in your changes are missing coverage. Please review.

Project coverage is 85.87%. Comparing base (0a8a3cc) to head (94a81ee).

Files	Patch %	Lines
crates/burn-jit/src/kernel/matmul/base.rs	62.19%	31 Missing ⚠️
.../burn-jit/src/codegen/dialect/gpu/vectorization.rs	0.00%	7 Missing ⚠️
crates/burn-jit/src/fusion/tracing/builder.rs	0.00%	5 Missing ⚠️
crates/burn-jit/src/kernel/matmul/tiling2d.rs	97.16%	4 Missing ⚠️
.../src/kernel/matmul/tiling2d_shader/write_output.rs	95.95%	4 Missing ⚠️
...t/src/kernel/matmul/tiling2d_shader/computation.rs	95.52%	3 Missing ⚠️
crates/burn-wgpu/src/compiler/wgsl/compiler.rs	81.25%	3 Missing ⚠️
...ernel/matmul/tiling2d_shader/load_shared_memory.rs	99.20%	2 Missing ⚠️
crates/burn-jit/src/kernel/matmul/tune/base.rs	87.50%	2 Missing ⚠️
...rates/burn-jit/src/codegen/dialect/gpu/variable.rs	50.00%	1 Missing ⚠️
... and 1 more

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1472      +/-   ##
==========================================
+ Coverage   85.79%   85.87%   +0.07%     
==========================================
  Files         657      660       +3     
  Lines       74022    74807     +785     
==========================================
+ Hits        63505    64238     +733     
- Misses      10517    10569      +52     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[nathanielsimard]

Only one comment, otherwise looks good

[nathanielsimard]

Can we extract those else where, the declaration is probably done in a for loop. Also, not sure we need to initialize val_0, val_1, val_2 and val_3 to zero.