[HW][SV] Allow procedural SV ops in hw::TriggeredOp

[fzi-hielscher]

Adapt the SV dialect ProceduralOp / NonProceduralOp traits to consider hw::TriggeredOp a procedural region. This should enable us to lower the procedural operations within a hw::TriggeredOp region before lowering the entire module.

See the discussion in #7292. We currently do not have a cross-dialect notion of procedural regions. However, the description of hw::TriggeredOp states bluntly that it creates a procedural region, so I think this is a relatively intuitive change.

Piggybacked are two sanity checks for hw::TriggeredOp:

• Don't allow nesting of TriggeredOps
• Ensure the number of input arguments matches the number of block arguments

[uenoku]

Hmm, I'm not sure it's reasonable change to regard regions hw::TriggeredOp as ProceduralRegion in verifier. I'm more comfortable if hw::TriggeredOp just implements ProceduralRegion trait in that case. In order to do that I think it's necessary to promote the notion of procedural regions from SV to HW but it will require more broader discussion.

[fzi-hielscher]

I agree that this is far from ideal. I had, in fact, tried to move ProceduralRegion into HW, but then stumbled over the question on how to deal with this transitively w.r.t. upstream dialects (notably SCF). CC #719.

So, I went with the quick and dirty solution allowing me to temporarily have a sv.if in a hw.triggered. 😬

[fzi-hielscher]

Superseded by #7335 .