This came up in discussions about #100376
See comment: #100376 (comment)

The change in the PR ended up handling BBJ_ALWAYS the same way it handles BBJ_THROW blocks, since it appears that some of them throw, but they most likely should be BBJ_THROW in the first place.

One possible reason for why this happens is that GenTreeCall::IsNoReturn() does not cover helper cases. However simple attempts to fix that ran into issues/asserts about throw helper merging, which was too unrelated to the PR.

Thus this issue - to follow up on:

• sort out the case when BBJ_ALWAYS may end with a call to athrowing helper
• sort out if GenTreeCall::IsNoReturn() could cover helper case as well.