Corrected tokens with LEMMA reflecting the CorrectForm, not the FORM

[rhdunn]

The following have a CorrectForm=inand are correctly marked up as Typo=Yes.

ERROR: Sentence n01104011 token 20 -- IN lemma 'in' does not match lowercase-form applied to form 'is', expected 'is'
ERROR: Sentence n01125009 token 15 -- IN lemma 'in' does not match lowercase-form applied to form 'is', expected 'is'

According to Wrong Morphology or Syntax the main fields should reflect the source text ("is") and not the corrected text ("in"). As such:

1. The LEMMA should be "is" if the surface POS is IN, but shoud be "be" if a verb.
2. A CorrectLemma=in should be added to reflect the corrected form.

I haven't confirmed this, but I suspect that the POS fields need to reflect that is is a verb, in which case it would need CorrectUPOS and CorrectXPOS per [1] as well.

[1] https://universaldependencies.org/misc.html#correctfeature

[AngledLuffa]

There are quite a few instances of lemmas having the corrected lemma in EWT, and I'm a little shy about deviating from the standard they set there

[rhdunn]

I'm planning on raising similar issues in the other treebanks. Maybe this is something that needs discussion in a docs issue?

My rationale for this issue is that it deviates from UD guidelines per the linked typo and MISC annotation docs.