I think splitting on reading will end up with two entries just as messy as this one.

koj, daijr, daijs, meikyo

i don't expect this change to be approved.  just showing what it would look like 
if we don't split by reading.  should we split?

@@ -7,0 +7,6 @@
+</k_ele>
+<k_ele>
+<keb>緩い</keb>
+</k_ele>
+<k_ele>
+<keb>微温い</keb>
@@ -14,0 +20,1 @@
+<re_restr>温い</re_restr>
@@ -16,0 +23,2 @@
+<stagk>温い</stagk>
+<stagk>微温い</stagk>
@@ -18,0 +27,1 @@
+<s_inf>ぬくい is primarily used in Western Japan</s_inf>
@@ -21,0 +31,16 @@
+<sense>
+<stagr>ぬるい</stagr>
+<pos>&adj-i;</pos>
+<xref type="see" seq="1214410">緩い・ゆるい</xref>
+<misc>&uk;</misc>
+<gloss>lenient</gloss>
+</sense>
+<sense>
+<stagr>ぬくい</stagr>
+<pos>&adj-i;</pos>
+<xref type="see" seq="2567890">温</xref>
+<misc>&uk;</misc>
+<misc>&arch;</misc>
+<gloss>slow</gloss>
+<gloss>stupid</gloss>
+</sense>