[perf] Wrap clear loss/grad into a kernel

[k-ye]

Let me submit this part first. With this simple tweak MPM lagrangian force on my end already runs from 15fps to 24fps...

Related issue = #N/A

[Click here for the format server]

---

[codecov]

Codecov Report

Merging #1902 into master will increase coverage by 0.98%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #1902      +/-   ##
==========================================
+ Coverage   42.87%   43.86%   +0.98%     
==========================================
  Files          45       45              
  Lines        6397     6190     -207     
  Branches     1099     1099              
==========================================
- Hits         2743     2715      -28     
+ Misses       3484     3306     -178     
+ Partials      170      169       -1     

Impacted Files	Coverage Δ
python/taichi/lang/__init__.py	42.46% <100.00%> (-0.51%)	⬇️
python/taichi/lang/meta.py	64.38% <100.00%> (+1.52%)	⬆️
python/taichi/lang/ast_checker.py	70.58% <0.00%> (-1.64%)	⬇️
python/taichi/testing.py	75.00% <0.00%> (-0.72%)	⬇️
python/taichi/lang/linalg.py	89.33% <0.00%> (-0.67%)	⬇️
python/taichi/misc/util.py	18.27% <0.00%> (-0.24%)	⬇️
python/taichi/misc/task.py	0.00% <0.00%> (ø)
python/taichi/tools/patterns.py	0.00% <0.00%> (ø)
... and 11 more

---

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 e0ac1d8...10fcb9c. Read the comment docs.

[archibate]

👍 for make Taichi.js export process a bit easier!