It seems that, after it became spirantal, an initial [ɣ] from [g] before a liquid sometimes became syllabic and became a vowel. This is distinct from its normal development, where it simply vanished. This rare development shows up only in a few, mostly archaic forms, as described in both the Outline of Phonetic Development (OP1) from the 1940s and the Outline of Phonology (OP2) from the 1950s:

In Q. dr, gr > r, gl, dl > l, but also ǝr, ǝl with ǝ redeveloping as full vowel appear, alaure beside laure, glaire [beside] laire (OP1: PE19/37).

The same result was also usual for gr, gl; but there are traces especially in AQ and in archaic forms used in poetic PQ of another development. It appears that at the stage ʒr, ʒl the ʒ sometimes became syllabic and vocalic, producing an indeterminate [a] that later became a full vowel of a quality decided by the vowel immediately following the l or r. Thus *glawarē "sheen of gold" > prim. Q glaurē, ʒlaurē > AQ, PQ laurē beside †alaurē (OP2: PE19/79).

Indications of this development can also be found in the Comparative Tables [CT] of phonetic development from the 1930s, where initial gr became ar or r, and initial gl became aly or ly (PE19/21). The second development is a bit strange and I can find no other sign of it (apparently initial gl > ly). In CT, Tolkien original wrote gl > Q[uenya] al, ly and L[indarin] al, l, the last of these matching the development in OP1 and OP2.

It seems that the syllabic spirant first became ǝ and then developed a vowel of the same character as the following vowel in the word. In addition to the AQ. alaurē example given above, other examples include:

• ✶grauk- > Q. arauko “demon” beside more common rauko (WJ/415).
• [deleted entry] ᴹ√GLINDI > ᴹQ. ilin “pale blue” (Ety/GLINDI).

[Hat tip to Lokyt for pointing out these two examples out to me.]

Conceptual Development: I can find no sign of similar syllabification of initial [ɣ] in Tolkien conception of Early Qenya in the 1910s and 1920s.