Improved NE scheme to address known issues

[khumnath]

improved some transliteration schemes.
this scheme can be used in all devnagari script based languages with some modifications.

[agx]

Tests pass now 👍

I think you can enable them again in .github/workflows/build.yml by removing the TestNe| there. This would make sure we don't accidentally regress.

[khumnath]

@agx The test will not pass with the current Ruby test file. The acceptCondition in Varnam does not handle multiple characters at this time. For example, in 'baagh' => 'बाघ', the sequence gh should return घ where gh is mapped to घ with :accept_if => :ends_with, so it should return घ but returns घ्.

[agx]

Ah, I see it too now:

  2) Failure:
TestNe#test_words [/var/scratch/librem5/varnam-schemes/test/ne.rb:26]:
Expected: "बाघ"
  Actual: "बाघ्"

Thanks @khumnath . Seems I hadn't rebuilt the vst properly before.