Termination positions for Frameshift/Nonesence SeqVars are not set properly

[gromdimon]

Describe the bug
There are examples, where pHGVS is not set (p.?), but do however exist in other databases.

To Reproduce
Example:
NM_004360.3:c.2506G>T (p.Glu836Ter) (Gene: CDH1)
It gives the pHGVS result by Autopvs1 as NP_004351.1:p.Glu836Ter (link: https://autopvs1.bgi.com/variant/hg19/16-68867259-G-T)
compare to one from mehari: 'NM_004360.5:p.?' (link: https://github.com/bihealth/autopvs1/blob/main/tests/assets/cdh1_mehari_seqvar.json)

Expected behavior
pHGVS should be set properly.

Screenshots
N/A

Additional context
N/A