Consider hiding sample files from language statistics

[rdipardo]

GitHub currently identifies this repository's primary programming language as Lean, because the one and only tracked file in that language has a disproportionately large byte count:

These stats are generated by github-linguist and locally reproducible:

$ cd userDefinedLanguages
$ github-linguist 

35.94%  45960      Lean
34.38%  43967      Python
9.80%   12531      Perl
7.77%   9934       LSL
3.32%   4250       Vim Script
2.69%   3436       Bicep
2.08%   2662       RobotFramework
1.76%   2248       E
0.80%   1019       Assembly
0.47%   601        Cython
0.47%   598        PowerShell
0.38%   489        YARA
0.16%   202        Max

They can be filtered by adding specially formatted rules to a .gitattributes file, e.g.,

# .gitattributes
# TODO: add EOL controls

UDL-samples/**    -linguist-detectable

After commiting this change, the actually used programming languages come into much clearer focus:

77.01%  43967      Python
21.95%  12531      Perl
1.05%   598        PowerShell

[pryrt]

It took a few refreshes (or maybe the checks had to finish first), but it eventually refreshed to: