Boosted block: possible era mismatch in `Point blk`

[tbagrel1]

Just after an era transition, the eligible block to receive votes/cert will be from the previous era. However Peras{Vote,Cert} and Peras architecture uses Point blk for convenience for the target of a vote/cert.

The blk in this corresponds to the era in which the parent datastructure lives, not really the era of the block being voted for/boosted.

Somewhat related to #246

[tbagrel1]

FTR, in Peras 25.5 we introduced Bytes32RealPoint that was supposed to act as an era-agnostic point for Peras{Vote,Cert}. However:

• We can't use it for blk = TestBlocks, whose HeaderHash is not always a 32-byte hash
• ChainDB uses Point blk everywhere
• In Peras 27, to guarantee type family injectivity, we phantom-tagged concrete V1.Peras{Vote,Cert} with blk despite our initial intent of making these monomorphic

Consequently, it seems simpler for now to use Point blk in concrete V1.Peras{Vote,Cert} too.

We may want to revisit this in the future.

[tbagrel1]

N.B. : After checking with @agustinmista , it seems possible that we vote/certify Origin, and it doesn't seem to be problematic. So having RealPoint instead of Point for Peras target might not be desirable in the end.