Improve sending of bls transaction by checkpointing module.

[KonradStaniec]

Currently at appropriate time checkpointing module just start concurrent goroutine which send bls transactions:
https://github.com/babylonchain/babylon/blob/main/x/checkpointing/abci.go#L45

This fine for now as it get job done.

The main downside is that if at epoch boundary, validator account would send another transaction concurrently (for example transfer) , then bls transaction could fail due to invalid sequence number error.

This is a bit corner case scenario as epoch won't change that ofter but taking into account that this is consensus critical tx it is worth improving this situation.

Probably simplest one is to add some kinda of retry mechanism.
Another one, is to track life time of bls transaction, and buffer transactions send from validator account through this node during this lifetime.

[gitferry]

Thanks for reporting this issue. Indeed, it is critical to consensus and should be improved carefully. @gusin13 has been working on a retry mechanism (#50 and #72) that has been applied in the vigilante reporter which also has the incorrect sequence number issue when sending BTC block headers to Babylon. As suggested by @vitsalis, maybe we can also use the same retry mechanism in the checkpointing module.

One bit is that I can't catch the invalid sequence number error as it is reported back in the raw_log of the tx response and it seems that the tx is successfully executed regardless of the error. @KonradStaniec do you have any ideas of how we can catch the error and do the retry?

[gitferry]

Resolved in #166