[CS2] Fix for comma after function glyph [#1043]

[helixbass]

The one "unambiguous" thing I saw in #1043 was that a , immediately following -> or => can indicate an empty function body, since otherwise it seems to be a syntax error (ie a function body can't start with a comma). So this uses the rewriter to interpret -> , and => , that way

@GeoffreyBooth I targeted 2, I guess this could safely go on master since the previous meaning was a syntax error, but you seem to prefer only targeting master if really necessary?

[helixbass]

@GeoffreyBooth actually it was super easy to lump in a fix for #3446 (which is what led me to #1043) / #3845. Same thinking applies: treat -> . or => . as an empty function body followed by chained whatever, and I believe here also the previous meaning would always be a syntax error. So now those chained calls with empty function args work eg:

a
.b ->
.c

So fixes #3446 and #3845, added tests with examples from those issues

[GeoffreyBooth]

This all looks good to me. @lydell?

[lydell]

Nice!

[connec]

This does also allow a -> .b (a(function () {}).b), which is a bit odd.

[helixbass]

@connec ya I'd thought there was no way to distinguish between "inline" chaining and line-break chaining since the leading . suppresses line tokens (eg INDENT/TERMINATOR) but looked more closely and the -> token will have .newLine set in the line-break case. So opened #4590 to distinguish between these so that eg a -> .b will still error