That addition of さ to 年かさ is not the problem; it's the greedy longest-match parser which would catch 年かさ anyway. I've added a pseudo entry for 年から, which should fix that case. let me know of others.

@@ -14 +14 @@
-<pos>&adj-na;</pos>
+<pos>&adj-no;</pos>

There is some spurious data hidden in this entry on account of which a さ is appended and the "Partial Match" comment added:
    年かささ from : 年嵩; 年かさ 【としかさ】 (adj-na,n) senior; older; elderly; ED [Partial Match!]
A second issue that i'd like to mention is that when using "Text Glossing", the occurrence of 年か - for example, within 年から - triggers the display of 年かささ - i think it would be "safe" to change the trigger to 年かさ, so that 年から no longer would no longer trigger this entry...

JWN, EDR
43k hits
Ordered by hits

@@ -6,0 +6,3 @@
+</k_ele>
+<k_ele>
+<keb>年かさ</keb>