I don't think that would be the only case where we base something on one sole attested example! As far as Sindarin goes, one attested form is outright luxury...

Actually, I think you are right to be skeptical. In the normal phonological development of Sindarin, primitive medial nt became nn. Compare S anna "give", Q. anta, G. antha, and S danna "fall", Q lanta, G lantha.

Also note that there are exceptions to this rule. Tolkien said that normally nt > nth > nh (voiceless double nn) but that the nth was preserved in spelling and sometimes restored, as in canthui "fourth" (VT42:27). In general, though, the result was nn (PE17:131, numerous examples).

So mentha-"send, cause to go" would be menna-, which is indistinguisable from menna-"go", and then clarified as mennada-, and then contracted as mentha- once again?

I think it is likelier that the Sindarin cognates of Q men- "to go" and menta- "to send" are S men- and menna-.

I never met for menna- for intransitive "to go" anywhere except Salo's movies. I think it's fair to accept men- for "go" and menna- for "send" (though this root is claimed to be characteristic for Quenya in PE17 with rare derivatives except i-ven "way"; thus men-, dadwen- could be Q. influence?)
Concerning your main question, I guess it's safe to render the mentioned verb as glinna-. I'm also suspicious of lth clusters though they are well-attested in the Etymologies, but supposed to give ll in "mature" Sindarin (And go nh, lh etc. while before a stressed vowel).