[bug] Fix matrix operations to use numpy only in Python scope

[k-ye]

is_pyconstant is not enough because there are cases where constant ti.Vectors are used in the taichi scope. For example:

taichi/examples/cornell_box.py

Line 37 in 6518a21

light_normal = ti.Vector([0.0, -1.0, 0.0])

This PR adds in_taichi_scope() and in_python_scope() helpers to make cornell box run again. As a slightly off the topic change, I also adopted the new gui.is_running to this example.

Related issue = #1051, #1190

• [Click here for the format server]
• [Click here for coverage report]

[rexwangcc]

Thanks for the fix @k-ye ! AFIK, there are 3 efforts now trying to resolve this issue: #1184 #1188 , since we are trying to fix it in the next release asap, we might want to come up with a plan on how to consolidate them and what is the short/long term plan for this.

Besides that, in_python_scope is better than what's used in #1184 !

[codecov]

Codecov Report

Merging #1208 into master will increase coverage by 0.22%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #1208      +/-   ##
==========================================
+ Coverage   66.85%   67.08%   +0.22%     
==========================================
  Files          35       35              
  Lines        4827     4827              
  Branches      885      885              
==========================================
+ Hits         3227     3238      +11     
+ Misses       1424     1414      -10     
+ Partials      176      175       -1     

Impacted Files	Coverage Δ
python/taichi/lang/matrix.py	84.79% <0.00%> (+1.85%)	⬆️

---

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 7562943...eb2578d. Read the comment docs.

[archibate]

Thank for fixing this! We may merge this for a temp solution in v0.6.9. But IMHO for long term consideration, v0.6.10, we should use #1188 as ultimate solution, which doesn't rely on to_numpy.

[yuanming-hu]

LGTM! Thanks!

[k-ye]

Thanks for the quick review, and sorry about the redundant works! I agree that we should do this systematically.

[rexwangcc]

LGTM! This PR now supersedes #1184

[archibate]

Thank for the quick solution, please don't forget to take a look at #1188 when you have time :)