Rename the region accessors for ReduceOp and ReduceWindowOp to match the intent and hlo spec

[sdasgup3]

The existing name body suits more the region name of whileOp

[sdasgup3]

@GleasonK
I think this change can have compatibility issues. Are we ready at this point in ascertaining the compatibility guarantees?

[burmako]

Another question is how strongly we feel about renaming these specific fields right away.

There are a lot of opportunities for improving various names in StableHLO/MHLO, but we've been holding off so far to avoid complicating the ongoing migration from MHLO to StableHLO, with the only exception of renaming operands to inputs because operands will stop working in a few weeks.

Perhaps we could start collecting renaming ideas in a ticket similar to #289 (@atondwal has an entire document with ideas), so that at a later point we could evaluate them all in concern and then execute on all of them at once to avoid slow dripping breaking changes.

[GleasonK]

@GleasonK
I think this change can have compatibility issues. Are we ready at this point in ascertaining the compatibility guarantees?

We don't offer API compatibility guarantees so I don't think this should be an issue. Can confirm by checking out main, serializing a file, and try to deserialize in this branch. Happy to help with any of that if needed.

[burmako]

As mentioned in the comment above, I'd like to propose to table the renaming discussions until later when we have a chance to review all the names. This way we'll ensure that: 1) the opset has uniform naming style, 2) all C++/Python breaking changes related to naming are done at once instead of spread over weeks or months of breakages.

[sdasgup3]

Superseded by #351

[sdasgup3]

@burmako Agreed and makes sense to me.
@GleasonK Thanks for the information! No action from you end at this point.