OK, closing.

https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kakuki/22/pdf/jitaihyo.pdf

As I understand it, the bureaucrats in the education ministry decided in 2000 that 搔 is the proper ("印刷標準") form of the kanji and 掻 is an acceptable simplified variant ("簡易慣用").

Practically every recently published kokugo from a variety of different publishers (daijirin, meikyo, iwakoku, etc.) uses these 印刷標準 kanji rather than the simplified forms.

Many entries will be much messier if we display both types of forms, so I think it's probably best to stick with one. Lately we've been hiding the "proper" forms and keeping the simplified forms visible. I think that's probably fine. If we want to switch to the 印刷標準 forms someday, it would be simple enough to write a script to swap them.

I agree [rK] probably isn't a good tag for these kanji forms, since they're quite a bit more common now than when the 2007 n-gram corpus was compiled. People are bound to encounter them frequently and wonder what they are. Maybe that information is better explained via a dedicated kanji dictionary rather than a vocabulary dictionary like JMdict.

掻く	89328
搔く	123

I don't think 搔く rates as [rK]. Kanjipedia is a form of kanwa; not the sort of reference we'd use to justify word level [rK] tags. 搔 entered the JIS standards in JIS X 0212 as an 異体字 of 掻. Even after 20+ years of Unicode it's still rather rare. Wiktionary says it's a 旧字体, but I don't take that as authorative.

@@ -10 +10 @@
-<ke_inf>&rK;</ke_inf>
+<ke_inf>&sK;</ke_inf>

https://www.kanjipedia.jp/kanji/0004265300
Not 旧字体, so [rK] rather than [oK]

@@ -10 +10 @@
-<ke_inf>&oK;</ke_inf>
+<ke_inf>&rK;</ke_inf>

@@ -25,0 +26 @@
+<misc>&uk;</misc>
@@ -30,0 +32 @@
+<misc>&uk;</misc>