[Pipeline] Add optional stall signal to pipelines

[mortbopet]

This signal is intended to connect to all stages within the pipeline, and is used to stall the entirety of the pipeline. It is lowering defined how stages choose to use this signal, although in the common case, a stall signal would typically connect to the clock-enable input of the stage-separating registers.

Future PRs will implement hardware lowerings - blocked by #5234.

[mortbopet]

CC @blakep-msft.

[teqdruid]

This is one style of stalling. It's my understanding that what we actually want is to stall only the stages which cannot proceed without clobbering the next stage. Meaning if the next stage doesn't have valid data in it, the current stage doesn't need to stall. @blakep-msft, is my understanding correct? If so, does this support that style?

[blakep-msft]

This is one style of stalling. It's my understanding that what we actually want is to stall only the stages which cannot proceed without clobbering the next stage. Meaning if the next stage doesn't have valid data in it, the current stage doesn't need to stall. @blakep-msft, is my understanding correct? If so, does this support that style?

I believe this PR just adds the stall input signal, which should be fine. The interesting cases will come up with the lowerings. If a pipeline has some stages which are stallable, and others which are not then you can have clobbering problems. One solution is to have stallable stages advance in spite of a high stall signal, in the case where the nearest upstream non-stallable stage has valid data in it. Representing this as a chain of pipelines (each pipeline in the chain can be stallable or not) may be a workable approach.

[teqdruid]

Blake makes a good point -- it's all in the lowering

[mortbopet]

That was also my view of the problem - a "stall this pipeline" signal seems generic enough to apply to all cases. Now, how that stall is actually implemented in hardware is a separate issue.