LArAthenaPool+LArTPCnv: TP separate LArTTL1 for writing RDOs in RNTuple (Strikes Back)

This MR T/P separates LArTTL1. This was originally implemented in !66964 (merged) for the main branch, where we T/P separated and schema evolved LArTTL1. However, the main issue is that we have esoteric workflows, e.g., Run 2 trigger simulation, which requires 21.0 and main to read/write each other's RDOs simultaneously. That means we have to teach both how to read the new type before writing it in main. So, the plan of attack is to:

• Merge this one into 21.0 and wait until we have a 21.0 nightly,
• Make sure everything works fine, then sweep the changes into main,
• Then eventually schema evolve main to _p1. At that point 21.0 should be able to read it, and in return produce an RDO main can digest.

Changes here also needs to go into the derivate 21.0 branches, e.g., 21.0-mc16a. I'll do this by-hand once we confirm all is OK.

Tagging @gemmeren, @mnowak, @jchapman, @ssnyder