x87: confusing use of float80_t for fbst

Subject: https://github.com/rfalke/decompiler-subjects/blob/master/from_holdec/i386_x87_fpu/ia32_elf/subject.exe
Output:
```
int32_t FBSTP(void) {
float80_t v1 = __asm_fbstp(-123456789011223344.0L); // 0x804a06b
*(float80_t *)&g2 = v1;
unsigned char v2 = *(char *)&g7; // 0x804a091
unsigned char v3 = *(char *)&g3; // 0x804a0b1
printf("FBSTP(%f) should=0x44 0x33 0x22 0x11 0x90 0x78 0x56 0x34 0x12 0x80 is=0x%02x 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x\n", -1.2345678901122334e+17, (int32_t)(int80_t)v1 % 256, (int32_t)v3, (int32_t)g4, (int32_t)g5, (int32_t)g6, (int32_t)v2, (int32_t)g8, (int32_t)g9, (int32_t)g10, (int32_t)g11);
return 0;
}
```

While the return type of `__asm_fbstp` is 10 bytes (80 bits) long it is not a floating point type. Therefor the type of v1 should be something else.

Start by inspecting the FBSTP output in issue 880 and the __asm_fbstp entry point shown there. Trace how its 10-byte return value becomes float80_t, then verify that the generated output uses a non-floating 80-bit representation while preserving the expected bytes.

Written by the indexing model from the issue text.