rm_stm: remove always true assert on transaction_ga feature

[bharathv]

Backports Required

• none - not a bug fix
• none - this is a backport
• none - issue does not exist in previous branches
• none - papercut/not impactful enough to backport
• v24.3.x
• v24.2.x
• v24.1.x

Release Notes

• none

[dotnwat]

why did we remove this assert with no explanation at all?

[bashtanov]

The explanation is it is always true. The feature is always available, as it depends on the code only and it has already been on for quite a few versions.

The background for the change is that feature_table and machinery around it misbehaves under some arguably unsupported scenario played in one of our tests. The scenario is having a node that both has empty storage and is not the highest version among all nodes in the cluster. The test is also under impression I'm not sure whether it is prohibited in the docs, and I wouldn't be surprised if it can happen in production. Say, a cluster is halfway through upgrade, and one node that still has the older version loses its disk. The test was also probably run with an incorrectly versioned build, see https://redpandadata.slack.com/archives/C06MCRY75U7/p1732741186471949

All in all, this change looks legit to me and it makes a test pass too, but seems to hide other problems in RP, tests and CI.