Disable legacy and v0 transaction formats

[apfitzge]

Problems

• Legacy and v0 transactions lack support of configuration in the formats and rely on compute-budget instructions and special runtime handling. This has a negative impact on validator performance.
• v0 transactions support address lookup tables which were a previous work around limited packet size, but add considerable complexity in the protocol/runtime.
• v1 transaction format addresses both of these problems but the older formats are still supported so the network has not yet realized the benefit of removing the supporting complexity

Proposal

Allow a migration period of no shorter than 180 days for users, developers, an apps to migrate to the txv1 format.
After the migration period, activate a feature to remove support in the validator for legacy formats in new blocks. RPC must continue to serve old formats

[topointon-jump]

Strong support for deprecating these formats! Just curious - in a previous discussion the idea of steadily increasing the CU cost of these was floated, why did we decide not to go that approach? How will we ensure this approach doesn't accidentally break people overnight?

[apfitzge]

This doesnt' necessarily mean we don't do that, this just sets the end goal that we are going to remove it.

Early adoption has actually been decent so far: https://dune.com/queries/8730642/12756092

If we get this accepted and still don't see the remaining folks adopting we could at that point consider the CU ramp approach.