Support more input types for categorical data.

[trivialfis]

• Shorten the type name from "categorical" to "c".
• Tests for np/cp array, scipy csr/csc/coo and uri.
• Specify the type for feature info.

Remaining type:

• pydatatable, xgboost will throw an error when categorical data is specified.

[codecov-commenter]

Codecov Report

Merging #7220 (7b629f3) into master (804b2ac) will decrease coverage by 0.02%.
The diff coverage is 87.87%.

@@            Coverage Diff             @@
##           master    #7220      +/-   ##
==========================================
- Coverage   82.63%   82.60%   -0.03%     
==========================================
  Files          13       13              
  Lines        4019     4024       +5     
==========================================
+ Hits         3321     3324       +3     
- Misses        698      700       +2     

Impacted Files	Coverage Δ
python-package/xgboost/data.py	69.64% <86.20%> (-0.10%)	⬇️
python-package/xgboost/core.py	84.35% <100.00%> (ø)

---

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 804b2ac...7b629f3. Read the comment docs.

[RAMitchell]

I don't understand what happens when I change the numerical types, is there more documentation somewhere?

[trivialfis]

Not at the moment, it's only used for text model dump. Actually, I have been thinking if it's possible to remove them along with the fmap parameter. (they are the same thing).

[trivialfis]

I removed the doc for numerical data. Let's just keep it as it's for now.

[trivialfis]

That's interesting (and unrelated to this PR):

[2021-09-15T11:52:48.240Z] tests/python-gpu/test_from_cudf.py::TestFromColumnar::test_cudf_training_simple_dmatrix terminate called after throwing an instance of 'thrust::system::system_error'

[2021-09-15T11:52:48.240Z]   what():  device free failed: cudaErrorLaunchTimeout: the launch timed out and was terminated

[2021-09-15T11:52:48.240Z] Fatal Python error: Aborted