Fix: #2389 Mempool walks, add try-mine CLI command

[kantai]

This PR addresses the most immediate mempool issue #2389, by modifying the walk logic to "try again" if there are mempool entries received at a particular height but not in the same fork.

This PR also adds a try-mine CLI command, as well as a diagnostic envar MEMPOOL_BAD_BEHAVIOR which executes the mempool walk in the manner that causes #2389.

This leaves unaddressed two potential issues:

1. Transactions received into the mempool at block A, but block A is "forked away". Those transactions will not appear in the miner's walk. However, such transactions will be viable for "replace-across-forks", such that if the transaction is rebroadcasted (not re-signed or fee upgraded), the miner will replace the existing tx in the mempool. (Issue: Mempool transactions accepted_at different forks are not considered #2394)
2. Mempool transactions are evaluated in "reverse" order: transactions received later are attempted first. This can impact chained transactions, because higher nonces will be tried first and fail, before the lower nonces are tried. In degenerate cases, this can lead to chained transactions only being included one-per-block. (Issue: Mempool transactions should be evaluated in nonce-order, not walk-order #2393)

I think #2389 is much worse than those two issues at the moment, so I'm not sure if we should try to address those issues at the same time, or instead split it into multiple releases.

[jcnelson]

LGTM, thanks for tackling this! Just one minor comment.

[lgalabru]

Looks good to me, thank you @kantai!

[blockstack-devops]

This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.