Fixing wrong cross-refs that are now incorrect deductions after changing mem regions

[fenugrec]

The disassembler is creating a wrong cross-ref due to an early mistake:

          000b16ec 20 79 00        movea.l    (DAT_0000ef0c).l,A0
          000b16f2 4e 90           jsr        (A0=>FUN_000081cc)

EF0C is in RAM, but earlier in my analysis I had some initialized data there, which contained the value 000081cc.
That RAM block is now correctly marked as "uninitialized", but I'm left with a large number of misleading refs like this.

Is there a way to undo the damage ? The obvious re-running the analysis doesn't change anything (it's probably for the best, that the analyzer is excessively careful not to delete refs , so no complaint here).

To Reproduce

• load attached .bin file as mc68000, big-endian, at address 0x88000
• don't analyze now
• Memory Map: add another region (read-only), starting at 0x8000, initialized with that same .bin file contents
• browse to b1682 and disassemble

Expected behavior
I'm not sure... Running analysis (one-shot or globally) should 'fix everything magically' ?
And disasm would look like

          000b16ec 20 79 00        movea.l    (DAT_0000ef0c).l,A0                                      = ??
                   00 ef 0c
          000b16f2 4e 90           jsr        (A0)

Screenshots
I noticed, in the 'wrong' db, if I open the Instruction Info window, the right-hand pane shows the unwanted address xref.

Attachments
Current state of the project:
https://filebin.net/35d94r4ip8i4v8ox/mc68_jsr.gzf
or, to start from scratch, the following .bin file must be loaded at address 0x88000 .
https://filebin.net/35d94r4ip8i4v8ox/U21_8000.bin.zip

Environment (please complete the following information):

• OS: archlinux
• Java Version: 21.0.6
• Ghidra Version: 11.3
• Ghidra Origin: distro package