That's a fair point. I'd misremembered how sankoku defined the red parentheses. I suspect uk is correct here because 嬰児 is more likely to be read えいじ.
In any case, I think unrelated readings like this are best kept separate if merging them means adding restrictions.

Google N-gram Corpus Counts
╭─ーーーー─┬────────┬───────╮
│ 嬰児　　 │ 21,061 │ 74.3% │
│ みどり児 │  2,395 │  8.4% │ - add
│ 緑児　　 │    340 │  1.2% │
│ みどりご │  4,557 │ 16.1% │
╰─ーーーー─┴────────┴───────╯

The appendix in sankoku defines the red colored parentheses as 「仮名書きにして（も）よい」

I'm not sure if it's okay to interpret that as "usually kana," which I think is a little stronger.

@@ -5,0 +6,3 @@
+</k_ele>
+<k_ele>
+<keb>みどり児</keb>

sankoku, smk:〔雅〕

Split from 1173630.