I've chatted with Tim about this and here is our conclusion:
Overwrite the existing epoch info in NWBv2 files, existing IPNWB functions: WriteSingleChannel, WriteEpochs, AppendEpochTable
Should be done only on request (i.e. just implement a function)
We will then do the epoch overwriting with an approach like MIES_MassExperimentProcessing.ipf
We can do that in another PR given that this PR is self-contained with tests.
Conceptionally it has to be solved that NWB2 export is threadsafe and epoch recreation is not.
An idea would be to iterate through all sweeps before and cache/buffer the recreated epochs such that NWB2 export can access the data threadsafe.
from https://github.com/AllenInstitute/MIES/pull/2041#issuecomment-2017914543
Conceptionally it has to be solved that NWB2 export is threadsafe and epoch recreation is not. An idea would be to iterate through all sweeps before and cache/buffer the recreated epochs such that NWB2 export can access the data threadsafe.