RFR: 8350896: Integer/Long.compress gets wrong type from CompressBitsNode::Value

[Emanuel Peter]

On Wed, 12 Mar 2025 08:21:21 GMT, Emanuel Peter <epeter at openjdk.org> wrote:

>> If I'm right here, you want to say how many many of the low significance bits are unknown, and that the rest, the high significance bits are all zero, right?
>
> You could say that we are trying to find case where we know that the compression result will have at least a certain number of leading zeros, which allows us to restrict the type.

It may also be more understandable if you used `min_leading_zeros` instead of `mask_max_bw`.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/23947#discussion_r1990934431