Test masp edge cases

[grarco]

We should try to test some masp edge cases:

• Test a block (or even batch) with one shielded action and one MASP tx in both orders and check that the shielded sync command always reconstruct the correct state
• Test a batch containing two masp txs, one successful and one failing (in both orders) and check that the protocol and the shielded sync command work as expected. Also test with an atomic batch (Masp tests #3840)

Also test some malicious masp transactions:

• A non-MASP transaction that carries a MASP section (check both the case as-is and the one where the tx manually triggers the MASP VP) (Masp tests #3840)
• A valid masp tx that carries more than one masp section (check that both the protocol and the shielded sync command ignore the unused masp sections) (Masp tests #3840)
• A valid shielded action tx that also carries a masp section (same checks as above)
• A MASP tx with an invalid conversion
• A MASP tx that does not reveal a necessary nullifier
• A MASP tx that does not append a required output description to the commitment tree
• A MASP tx that arbitrarily rewrites the commitment tree
• A MASP tx with a mismatch between the sapling value balance and the actual changes to the transparent balance
• A malleability attack on an outgoing shielded action where the actual IBC target gets modified
• An outgoing shielded action without the required masp section
• An outgoing shielded action with a wrong masp section
• An incoming shielded action with no masp transaction in its memo