Consistent coding for the naming of LR workers

Hi,

There are different types of Logical Replication workers -- e.g.
tablesync workers, apply workers, and parallel apply workers.

The logging and errors often name these worker types, but during a
recent code review, I noticed some inconsistency in the way this is
done:
a) there is a common function get_worker_name() to return the name for
the worker type,  -- OR --
b) the worker name is just hardcoded in the message/error

I think it is not ideal to cut/paste the same hardwired strings over
and over. IMO it just introduces an unnecessary risk of subtle naming
differences creeping in.

~~

It is better to have a *single* point where these worker names are
defined, so then all output uses identical LR worker nomenclature.

PSA a small patch to modify the code accordingly. This is not intended
to be a functional change - just a code cleanup.

Thoughts?

------
Kind Regards,
Peter Smith.
Fujitsu Australia