I thought this one was very disappointing indeed. Similarly to x_word_fan, I managed to complete the grid without finding most of the answers clued without definition, as there was simply no need to do so. Once one knew what was going on, the thematic answers just slotted in.
This seemed like very lazy setting to me; I'd have much preferred to have fewer thematic entries, but 'proper' ones where both the initial entries and then the replacements could fit in the same slots, both giving equally valid words, if that makes sense. As it was, it seems the setter has simply found X replacements (from a very large set) that fit in the grid, and then found X initial entries which could, frankly, have slotted in anywhere.
After a long run of very good puzzles in the EV, this was a definite drop in standard.