restore getfield deprecation for foo.(1) etc.

[stevengj]

This restores the deprecation warning for object.(i) when i is a literal integer, which was accidentally broken by #17318 #17300 as noticed by @tkelman.

(I'm not sure how to do a regression test for a deprecation warning. 😕)

[tkelman]

Thanks! BTW, are the M.(:+) and M.(1) deprecations mentioned in NEWS?

edit: M.(:+) is, but M.(1) isn't from what I could find while doing #17779

[stevengj]

Yeah, probably M.(1) should be mentioned as well.

[stevengj]

I fixed the NEWS for M.(1) in 47d522f.

[stevengj]

Okay to merge?

[tkelman]

Looks reasonable to me, but what the rest of the compress-fuse (that isn't being changed) accomplishes goes a little bit over my head. If Jeff approves?

[stevengj]

cc @JeffBezanson