Most of these seem correct to me.
Regarding ara + gûr - soft mutation should apply; this would reduce the whole thing to... Aror/Arur? Not the best I suppose... I am guessing you were looking at the names of the Chieftains of the Dúnedain, which seem to sporadically ignore this mutation; I don't know how applicable this approach is to proper Elvish.
Regarding Angmund and Angvir:
Both are tricky.
The examples of mund in compounds come from a 1972 letter, i.e. from a period where Tolkien is known to have been very forgetful at times. The forms he gives in that letter are Aramund (‘Kingly bull’), Tarmund (‘Noble
bull’), Rasmund (‘Horned bull’), Turcomund (‘Chief of bulls’); of these, Turcomund clearly seems to start from a Quenya element... In fact, if we were to adapt them as Aramundo, Tarmundo, Turcomundo, they'd be perfectly acceptable and sensible Quenya names. Rasmund is an outlier in this regard, since -sm- is not otherwise attested in either Quenya or Sindarin; it could, perhaps, be a correctly constructed Sindarin form, but who knows (in Q. it would be Rassemundo).
In other words all of the forms in that letter are somewhat puzzling, and given the context of Tolkien's forgetfulness in his old age, I would personally recommend to not take them as examples of anything (except, maybe, of -ssm- > -sm- being possible in Sindarin).
With this in mind, I would say that ang + mund and ang + mîr should end up with the same sequence of consonants in the middle. But what sequence?
-ngv- is attested once, in Gnomish (=the earliest drafts of Sindarin; specifically this attestation from 1917), which should not be applicable to Sindarin. Meanwhile -ngm- is also attested in exactly one word/name (though more than once naturally) - Angmar; except here we also have Tolkien saying:
"[...] Angmar seems to be corrupt Sindarin or Quenya < anga "iron" + mbar- "habitation": pure Sindarin would be angbar, Quenya angamar." [From the manuscript of Nomenclature of Middle-earth, in A Reader's Companion p. 20]
Which means that (a) this instance comes from [ᵐb] rather than [m], and (b) there is no true attested case of -ngm- in Sindarin.
So both options for these compounds are out of the window as far as attestations go... The best advice would be to avoid them altogether. If you must, though, I would hesitantly guess Angmund, Angmir to be the likelier outcome - both with a triconsonantal [ŋgm] cluster.