Writing to/from csv truncates joint names

[DavidLSmyth]

bvh-toolbox/src/bvhtoolbox/convert/bvh2csv.py

Line 151 in 50d62b7

data = np.array(data, dtype=[('joint', np.unicode_, 20),

bvh-toolbox/src/bvhtoolbox/convert/bvh2csv.py

Line 152 in 50d62b7

('parent', np.unicode_, 20),

limits the length of the joint names to 20, but in tests, you have joint names which are much longer, creating artifacts in the csvs such as:

mixamorig:LeftHandMi,mixamorig:LeftHandMi, 3.60930, -0.16690, 0.07780
mixamorig:LeftHandMi,mixamorig:LeftHandMi, 3.40040, -0.62740, 0.11480
mixamorig:LeftHandMi,mixamorig:LeftHandMi, 3.62050, -0.64930, 0.11990

Maybe adding an invertibility test (bvh -> csv -> bvh) would be of value? This could be part of #11

[OlafHaag]

Thanks for reporting the issue! I'm not sure why I did limit it to 20 characters. I guess, it's a relic from the data I originally developed this module for. If you need a fix asap I could just raise the limit to 120 or whatever, otherwise I plan to revisit this whole package at the end of this year. There's a lot I would like to refactor and as you mentioned, adding tests.

[DavidLSmyth]

No worries, as others have pointed out in other issues, the csv to bvh feature is a handy one as more datasets are being released which contain pose data as csvs instead of bvh or other formats. No need for an immediate fix on my end, thought better to flag for anyone else using. Thanks for the great work!