`filewriter`: selog `ICP_DAE_TD` and `ICP_SYS_TD`

The existing `.nxs` files always contain these datasets, which describe time differences between the DAE and computer clocks.

These differences do not exist in the same way in the streaming system. Therefore we need to decide what we do with these datasets.

## Potential differences from existing files

I am proposing to **not** write these datasets, as their meaning is sufficiently different in the streaming system that these names are confusing. If downstream consumers are using these datasets, I would like to understand why, as the assumptions may have been broken in streaming system.

Start at the filewriter entry point and inspect how streaming handles the ICP_DAE_TD and ICP_SYS_TD datasets compared with existing .nxs files. Check whether downstream consumers rely on these datasets; done means the differing streaming semantics and the decision about writing them are documented and agreed.

Written by the indexing model from the issue text.