x87: incorrect implementation of fprem1

Subject: https://github.com/rfalke/decompiler-subjects/blob/master/from_holdec/i386_x87_fpu/ia32_elf/subject.exe
Output:
```
int32_t FPREM(int32_t a1) {
// 0x804ab70
printf("FPREM(11 mod 7): should= 4.000000/ 4.000000e+00 is=%12f/%15e\n", 4.0, 4.0);
printf("FPREM1(11 mod 7): should= -3.000000/ -3.000000e+00 is=%12f/%15e\n", 4.0, 4.0);
...
```

There are cases where the semantic of fprem and fprem1 differs. Like in the above case.

Start with the linked i386_x87_fpu/ia32_elf/subject.exe case and reproduce the FPREM/FPREM1 output shown in the issue. Trace the decompiler's x87 handling for FPREM1 and compare its result with the documented distinction, considering the work done when FPREM1(11 mod 7) reports -3.000000 rather than 4.000000.

Written by the indexing model from the issue text.