Fix equal-always? on fl/fxvector, stencil-vector

[AlexKnauth]

Checklist

• Bugfix
• tests included

Description of change

Fixes equal-always? and chaperone-of? on flvector, fxvector, and stencil-vector, to return false on different values with the same contents at the time, because they are mutable.

Fixes #4983.

• equality tests
• hash-code tests
• Racket CS
• Racket BC

[mflatt]

Looks ok to me. I worry a little that the fallthrough to #%equal? has changed, or maybe I'm misreading; even if I'm reading right, I can't see any reason that the change would create a problem.

[AlexKnauth]

Yes, the fallthrough to #%equal? has changed in that it only falls through to that when the mode is equal? or impersonator-of?, not equal-always? or chaperone-of?. Are there any more immutable data structures that would need that fall through that I haven't accounted for?

[mflatt]

There are no more that I can think of.

In trying some things, I discovered that paths are currently not handled right for equal-always?. The repairs (which I have ready to push later) are independent of your changes. If there's something else wrong like that on CS, I expect the problem will be again with the struct and not with equal?.