consider updated logic for check_tr_trdup

[sarabodach]

From @harriscw via https://github.com/pharmaverse/sdtmchecks/pull/281#issue-2051488428:

@sarabodach this is a PR into your PR. Here's some thoughts:

• Including TRSPID in the dup variables seems sub optimal - would it only catch a mapping issue or something? But also trying to add RAVE in at the end also isn't perfect - by just merging on variables a guess is made of which form the dup is in. And the guess is not always right (e.g. the same info can be in indicator vs non indicator form). Ultimately I decided to leave SPID/RAVE out of this check.
• Duping with VISIT seems to get rid of a lot of records where date duplicate is flagged due to missing date. While missing dates may be bad its not technically a duplicate if its at different visits. Checking for missing dates could be a separate check.
• TRGRPID seems needed in the dup vars as well

[sarabodach]

If check_tr_dup is removed, can consider using lifecycle::deprecate_warn()