Deprecate all pyfontaine tests

[m4rc1e]

In check-ttf we had the --no-coverage flag to disable pyfontaine, #1345. This came about because the implementation was was too slow.

I don't want to disable pyfontaine, I just want cleaner output.

The current output is very bad for QAing families containing many styles.

>> com.google.fonts/test/132 with ((u'font[0]', '/Users/marc/Documents/googlefonts/fonts/ofl/somesans/SomeSans-Bold.ttf'),)
   Checking Cyrillic Historical glyph coverage.
 * FAIL: Pyfontaine output follows:

Fonts
  Font:
    Filename: SomeFont-Bold.ttf
    Common name: Some Font
    Sub family: Bold
    Weight: bold
    Fixed width: yes
    Fixed sizes: yes
    Copyright: Copyright 2017 Google Inc. All Rights Reserved.
    Version: Version 1.101
    Glyph count: 785
    Character count: 732
    Orthographies:
        Orthography:
          Glyphs in set: 446
          Glyphs in font: 407
          Percent coverage: 91
          Missing values: 
              U+0460  Ѡ  CYRILLIC CAPITAL LETTER OMEGA
              U+0461  ѡ  CYRILLIC SMALL LETTER OMEGA
              U+0464  Ѥ  CYRILLIC CAPITAL LETTER IOTIFIED E
              U+0465  ѥ  CYRILLIC SMALL LETTER IOTIFIED E
              U+0466  Ѧ  CYRILLIC CAPITAL LETTER LITTLE YUS
              U+0467  ѧ  CYRILLIC SMALL LETTER LITTLE YUS
              U+0468  Ѩ  CYRILLIC CAPITAL LETTER IOTIFIED LITTLE YUS
              U+0469  ѩ  CYRILLIC SMALL LETTER IOTIFIED LITTLE YUS
              U+046c  Ѭ  CYRILLIC CAPITAL LETTER IOTIFIED BIG YUS
              U+046d  ѭ  CYRILLIC SMALL LETTER IOTIFIED BIG YUS
              U+046e  Ѯ  CYRILLIC CAPITAL LETTER KSI
              U+046f  ѯ  CYRILLIC SMALL LETTER KSI
              U+0470  Ѱ  CYRILLIC CAPITAL LETTER PSI
              U+0471  ѱ  CYRILLIC SMALL LETTER PSI
              U+0476  Ѷ  CYRILLIC CAPITAL LETTER IZHITSA WITH DOUBLE GRAVE ACCENT
              U+0477  ѷ  CYRILLIC SMALL LETTER IZHITSA WITH DOUBLE GRAVE ACCENT
              U+0478  Ѹ  CYRILLIC CAPITAL LETTER UK
              U+0479  ѹ  CYRILLIC SMALL LETTER UK
              U+047a  Ѻ  CYRILLIC CAPITAL LETTER ROUND OMEGA
              U+047b  ѻ  CYRILLIC SMALL LETTER ROUND OMEGA
              U+047c  Ѽ  CYRILLIC CAPITAL LETTER OMEGA WITH TITLO
              U+047d  ѽ  CYRILLIC SMALL LETTER OMEGA WITH TITLO
              U+047e  Ѿ  CYRILLIC CAPITAL LETTER OT
              U+047f  ѿ  CYRILLIC SMALL LETTER OT
              U+0480  Ҁ  CYRILLIC CAPITAL LETTER KOPPA
              U+0481  ҁ  CYRILLIC SMALL LETTER KOPPA
              U+0482  ҂  CYRILLIC THOUSANDS SIGN
              U+0483  ҃  COMBINING CYRILLIC TITLO
              U+0484  ҄  COMBINING CYRILLIC PALATALIZATION
              U+0485  ҅  COMBINING CYRILLIC DASIA PNEUMATA
              U+0486  ҆  COMBINING CYRILLIC PSILI PNEUMATA
              U+0487  ҇  COMBINING CYRILLIC POKRYTIE
              U+0488  ҈  COMBINING CYRILLIC HUNDRED THOUSANDS SIGN
              U+0489  ҉  COMBINING CYRILLIC MILLIONS SIGN
              U+a64c  Ꙍ  CYRILLIC CAPITAL LETTER BROAD OMEGA
              U+a64d  ꙍ  CYRILLIC SMALL LETTER BROAD OMEGA
              U+e0ff  
              U+f000  
              U+effd  
          Common name: Google Fonts: Cyrillic Historical
          Support level: partial

   Result: FAIL

The check is run several times on each font for different codepages. If I save a log, pyfontaine error messages make up over 80%. It makes FB almost impossible to use.

[graphicore]

I had an idea recently to make this generally better:

1. We should know which encodings are expected to be in a font. Can we use METADATA.pb subsets for this?
2. if an encoding is expected and pyfontaine passes it's a PASS
3. if an encoding is expected and pyfontaine fails it's a FAIL and the default error message is quite helpful.
4. if an encoding is not expected and pyfontaine fails, we make it a PASS
5. if an encoding is not expected and pyfontaine passes, we make it a FAIL. This requires to register the missing subset. Could be a WARN as well.

[graphicore]

To worst thing about the pyfontaine checks is that most of them fail by default. The above would regulate that. Also, once a font is released we should know which encodings are supported. The above adds a nice regression detection.

[davelab6]

I don't want to disable pyfontaine, I just want cleaner output.

I think we should turn pyfontaine off completely for now, it is a standalone tool and plugging into it quickly was just a stop gap.

The real need for now is to implement #1466