Restart dropped elections for unconfirmed blocks with higher PoW

[guilhermelawless]

With this and #2306 prioritization should be fully guaranteed.

Requires:

• block get to check for work
• account get + confirmation height get to check for confirmation

A 60s buffer of dropped elections is added which sets the maximum time another node has to send the upgraded block before it will only be confirmed via frontier confirmation. This was added to prevent high DB usage for any old block, instead it is only done if the root has been recently dropped. This grants protection from bad actors as they do not know which elections we dropped.

However:

• It is possible to infer with a certain probability which blocks each node is dropping if they use a default config
• It is not ideal to have a maximum time to upgrade a block

Therefore, another approach will follow in the future using a large container of passive elections.

[guilhermelawless]

Marked as incomplete, will add a cache of recently dropped elections to reduce DB I/O in this path.

[guilhermelawless]

Finished, up to 32k dropped elections (less than 100 bytes each, 3MB) and only considered for checking the DB if the upgraded PoW arrived within 60 seconds (can increase this value).

[cryptocode]

LGTM, just a minor refactoring suggestion