new ExtPos feature on fixed expressions

[AngledLuffa]

There's a new ExtPos feature on fixed expressions which triggers a whole bunch of validator warnings. There are also some actual errors now that it is expected that fixed expressions have this marked on them to account for unexpected POS for a particular dependency type. such as, for example, gets annotated as follows now in GUM:

31      such    such    ADJ     JJ      ExtPos=ADP      37      case    37:case _
32      as      as      ADP     IN      _       31      fixed   31:fixed        _
33      web     web     NOUN    NN      Number=Sing     35      compound        35:compound     Entity=(83-abstract-new-cf6-5-sgl|SpaceAfter=No|XML=<w>
34      -       -       PUNCT   HYPH    _       33      punct   33:punct        SpaceAfter=No
35      based   base    VERB    VBN     Tense=Past|VerbForm=Part|Voice=Pass     37      amod    37:amod MSeg=bas-ed|XML=</w>
36      faculty faculty NOUN    NN      Number=Sing     37      compound        37:compound     Entity=(84-person-new-cf8-1-sgl)
37      directories     directory       NOUN    NNS     Number=Plur     29      nmod    29:nmod:such_as MSeg=directorie-s

Is this a feature we can work on integrating into ParTUT? @nschneid @amir-zeldes

[AngledLuffa]

for reference:

https://quest.ms.mff.cuni.cz/udvalidator/cgi-bin/unidep/validation-report.pl?UD_English-ParTUT

[nschneid]

Explanation: https://universaldependencies.org/changes.html#extpos-for-fixed-expressions

Basically if a fixed expression attaches as case, it should be ExtPos=ADP; if it attaches as mark, ExtPos=SCONJ; if it attaches as cc, ExtPos=CCONJ; and so on.