In that, case, here's my idea:
The assumption here is that each footnote is first marked by an asterisk where it should be in the text, followed by some text, followed by an asterisk and the actual footnote. I assume, as in your example, that the footnote itself contains no markup- you'd need to adapt accordingly if that varies for other books.
In the search & replace conversion setting, use the search pattern
Code:
(?s)\*(?P<text>[^*]*?)*(?P<footnote>[^<]*?)
and the replacement string
Code:
\g<footnote> \g<text>
The caveat is that I haven't tested it. When you test the regular expression in the wizard, the text from the first asterisk to the end of the footnote should be highlighted.