The crux of Elaran's argument against ennas, as far as I understand it from my own conversations with him, is that the King's Letter is a relatively early text which does not neatly fall into the conceptual stage of "Sindarin", but is rather more representative of late "Noldorin" (i.e. the conceptual precursor of Sindarin), or at most a transitionary stage betweent the two.
Concepts conceived at the Noldorin conceptual stage can often be adapted to fit Sindarin (e.g. Noldorin auth "war" cannot work in Sindarin due to differences in phonological evolution between Noldorin and Sindarin, but it can be easily updated to oeth, which does work in Sindarin). However, not everything can be adapted/updated as easily.
Elaran's argument against ennas here is that it seems to be the sole attestation of the root √EN with the sense "yonder, over there" in the Sindarin branch of Elvish; as such, since there are no attestations of this root in conceptually later Sindarin, Elaran argues that this root cannot be applied to Sindarin (or at least it is not safe to assume so), and should be treated as a Quenya-only root.
A further argument he raised is that ennas, being likely derived from a primitive form ✶entassē, itself likely containing the strengthened form of the locative preposition ✶sē, is the only attested example of the strengthened form of ✶sē in the Sindarin branch of Elvish; thus, again, he argues that it is inapplicable in the later conceptual stage of Sindarin.
Having said all of this, two things must be pointed out: (1) Tolkien never, as far as I know, explicitly rejected any of this himself, this is merely the result of an analysis of the data; (2) there is no strong consensus on the matter as such within the linguistic community (of which I am a member myself).
Though I will not deny that Elaran's argument is very much valid, I myself disagree with it. It boils down to a difference in approach: I see no issue with reflexes of √EN in Sindarin, seeing as Tolkien did at some (not very early) point apply this root to both languages, and never (as far as I know) explicitly reject that attestation nor explicitly contradict it or state that the root is Quenya-only.
As for proposed alternatives, ta/to/taw as reflexes of √TA can work for "then, that, there", and taw is also attested as "thither" (and as "wood"). These work well enough, and I use them myself (interchangeably with derivatives of √EN). But I would not say they are definitive replacements of ennas, because I do not think a replacement is definitevely warranted in the first place.
