use a target language as source for translation memory

or legacy software is using a proprietary localization service.

file looks like:

@MlgKeySomething
ENG: english translation
GER: german translation

we have converted those files to pot and po and are importing them into zanata
we are using MlgKeySomething as the source string
so our pot contains the MlgKeySomething
en subfolder contains the po files with english translations etc..
so far everythig went smooth.

now we found an issue with TM:
it uses the source string as the matching criteria for suggestions. and as this is in our case an (hopefully) unique key, this is pretty useless

our new webui related stuff uses the standard and is working fine, but tons of legacy server side translations will not work in TM

so the feature request would be:
allow to set a translation i.e. en as the source languange for TM matching

i am also fine with any feasible workaound proposals.