Looking at 4 bytes at 34 byte offset from the start of the page for the innochecksum'ed files should work. It could be done a few times per tablespace to confirm (offsets 0+34, 16384+34, ...)
But in any case the assertion errors have not indicated so far that any pages have corrupted checksums. Thus, if there's corruption there, it wouldn't be diagnosed by innochecksum, and it is possibly a result of an unknown server bug.
Looking at 4 bytes at 34 byte offset from the start of the page for the innochecksum'ed files should work. It could be done a few times per tablespace to confirm (offsets 0+34, 16384+34, ...)
But in any case the assertion errors have not indicated so far that any pages have corrupted checksums. Thus, if there's corruption there, it wouldn't be diagnosed by innochecksum, and it is possibly a result of an unknown server bug.