feat: use LoopVectorization for faster operations

[avik-pal]

I'm getting significant speedups from this. Zygote pretty much works OOTB because we already have the rrules defined. SciML/NeuralOperators.jl#17 should be pretty much handled once this is merged.

fixes #94

TODOs

• Replace all loops with @tturbo.
  • activation gradient loop not working see shortcomings
  • replace some of the reduction operations with VectorizedReductions.jl
• Introduce a fast_activation that is completely OOP but still allows for LoopVectorization optimizations
• matmul and matmuladd
  • polyalgorithm that uses LV for smaller matrices
  • handle internal_operation_mode correctly
  • add CRC.rrules
• Enzyme Support
  • matmul
  • activation
  • affine_normalize
    • groupnorm
    • batchnorm
  • bias_activation
  • dropout
  • normalization -- update statistics. Can we mark it non-diff for enzyme?
  • fast_mean_var
• Downstream Errors
  • Dimension checking in matmul and matmuladd
  • Enzyme tests run locally

Current Shortcomings

• We can't @turbo the activation gradient function. We might want to hardcode some of the common cases for gradient computation -- tanh, gelu, sigmoid and relu would be my top contenders here

[codecov]

Codecov Report

Attention: Patch coverage is 50.36855% with 202 lines in your changes missing coverage. Please review.

Project coverage is 73.86%. Comparing base (ffffa89) to head (eef1dc0).

Files	Patch %	Lines
src/impl/affine_normalize.jl	40.12%	94 Missing ⚠️
src/impl/dropout.jl	35.55%	29 Missing ⚠️
src/impl/bias_activation.jl	42.55%	27 Missing ⚠️
src/impl/matmul.jl	73.68%	25 Missing ⚠️
src/impl/activation.jl	32.00%	17 Missing ⚠️
src/utils.jl	33.33%	6 Missing ⚠️
src/impl/fused_dense.jl	86.95%	3 Missing ⚠️
src/impl/normalization.jl	75.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #111      +/-   ##
==========================================
- Coverage   80.58%   73.86%   -6.73%     
==========================================
  Files          29       30       +1     
  Lines        1329     1580     +251     
==========================================
+ Hits         1071     1167      +96     
- Misses        258      413     +155     

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