Incorrect LLIL and Data and Variable References

Version and Platform (required):

• Binary Ninja Version: 5.3.8857-dev Ultimate, 70683f1f
• OS: macos
• OS Version: 26.2.0
• CPU Architecture: arm64

Bug Description:
I'm not sure exactly what is happening with this function, but there are a few analysis results that appear off. The LLIL is:

   0 @ 180070190  <return> tailcall(rax)

But should it be: tailcall([rip+disp]) or tailcall([rip+0x1062])?

In addition to the LLIL oddness, there is a false data reference being listed as 0x62 ??:

And then there are a large number of odd variable references:

And there is an error reported by BinExport:

[0 Default] Expected 16 bytes at 180070190, got 6

Steps To Reproduce:

1. Open this raw file: scalar key designs gracefully
2. Navigate to 0x180070190
3. Look for problems listed above.

Expected Behavior:
tailcall([rip+disp]) or tailcall([rip+0x1062]) in LLIL (or what you think is appropriate if I'm wrong) and the odd references not generated.

Binary:
scalar key designs gracefully

Additional Information:
The above issue's odd cross ref can be used to trigger this other GUI bug: https://github.com/Vector35/binaryninja-api/issues/7593

Open the raw file "scalar key designs gracefully" and navigate to 0x180070190; inspect the LLIL, data reference, variable references, and BinExport error shown in the report. Compare the observed analysis with the expected indirect tailcall and confirm that the false references and BinExport size error are resolved.