Benchmarks: Add Feature - Add option to use fp32 instead of tf32

[guoshzhao]

Description
Add option force_fp32 to use fp32 instead of tf32, only takes effect on Ampere or newer GPUs.

[codecov]

Codecov Report

Merging #213 (18ae97c) into main (dfbd70b) will decrease coverage by 0.00%.
The diff coverage is 87.50%.

@@            Coverage Diff             @@
##             main     #213      +/-   ##
==========================================
- Coverage   88.72%   88.72%   -0.01%     
==========================================
  Files          58       58              
  Lines        2821     2829       +8     
==========================================
+ Hits         2503     2510       +7     
- Misses        318      319       +1     

Flag	Coverage Δ
cpu-unit-test	74.46% <85.71%> (+0.02%)	⬆️
cuda-unit-test	88.65% <87.50%> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
...perbench/benchmarks/model_benchmarks/model_base.py	85.52% <80.00%> (-0.19%)	⬇️
...rbench/benchmarks/model_benchmarks/pytorch_base.py	75.53% <100.00%> (+0.80%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update dfbd70b...18ae97c. Read the comment docs.