Looks like this breaks the pull tester, which relies on BIP34 switch-over logic as it was.

tested ACK

@jgarzik Thanks for testing

@TheBlueMatt can you take a look? It looks like we'll have to change some assumptions in the comparson tool.

Here's a version rebased on top of #5968 and adding so more consensus params to avoid that ContextualCheckBlockHeader has to have a new couple of lines every time that a softfork gets old enough to enforce it by height rather than voting:

https://github.com/bitcoin/bitcoin/compare/master...jtimon:bip34_switchover?expand=1

Minor bug*: If the blockchain were reorganised off the switchover block, we would be breaking consensus here. I wonder if it might be good to abstract #5927 so its logic is reusable for things like this. That is, instead of "when do we enforce BIP34?", we ask "what blocks are permitted to skip BIP34 validation?"

* This is only a bug when checkpoints are disabled or #5927 is merged.

Closing this. It has no priority at all. I suppose this could be taken as part of the new bit-based softfork tracking system, at least when the old system is removed.