cumulus: add segment submission primitive types

[iulianbarbu]

Description

Adds collation-generation subsystem types that are used by the collators to submit segments. Currently the new types have corresponding types in cumulus, to enable the support for segment submission from the block_builder/collation_task levels.

We clearly separate now between V2 and V3/V4 messages - V2 submits CollatorMessage (named as before, although I would've changed it, and without the scheduling_proof, which was set to None either way), while V3 submits CollatorResubmitSegment, with various info that's used by the collation-task to craft the correct SubmitSegment collation-generation message.

The current block building/collation-task changes keep the existing submission contract for V2/V3. We attempt to build bundles for each core (for both V2/V3), but for V3, depending on whether it builds a bundle for the first core or not, it either sends across to the collation-task a CollatorResubmitSegment with a SegmentKind::WithBundle (appending a newly built bundle to an existing list of unincluded blocks, computed in a future resubmission manager, and submitting that - for now, the list contains just the newly built bundle) or a SegmentKind::ResubmitOnly (that's concerned just with resubmitting an existing unincluded segment as is, again computed by a future resubmission manager, and at the moment being handled as a no-op).

Closes #12310.

Integration

N/A - segment submission primitive types usage is gated by the CandidateReceiptV3 node feature on the relay chain side, by the VerifySignedSchedulingInfo parachain runtime config type (for which we do not have yet an implementation, pending on #12097), but it also lacks the requried collator resubmissions machinery (wip here: #12228, depending on this PR) or the V4 collator-protocol (wip here: #12172 for the collator-side, and a validator-side PR that's not yet up).

Review Notes

Foundation for the resubmission machinery, and for the V4 collator-protocol's collator-side concerns, of distributing a whole segment.

TODO:

• I am in the process of rebasing cumulus: unincluded segment submission with every RC slot #12228 on top of this work, some rough edges came up, and will mark this as ready to review once those are polished.