Regression: compare got harder to understand

Subject: https://github.com/rfalke/decompiler-subjects/tree/master/from_holdec/dmi/exp/ia32_elf/subject.exe

Before
```
int32_t intermediate_2_cmp_with_extra(int32_t a1) {
// 0x8048551
if (a1 <= 19 && a1 > 10) {
```

Now
```
int32_t intermediate_2_cmp_with_extra(int32_t a1) {
if ((uint32_t)(a1 - 11) <= 8) {
```
This "subtract and compare it as unsigned" is a nice implementation trick but decreased the readability.

Reproduce the output for the linked subject.exe and compare the Before and Now examples in the issue. Trace the decompiler path that emits the unsigned subtract-and-compare form, then determine how the result should be rendered so the original bounds are readable. Done means the generated comparison is clearer without regressing equivalent behavior.

Written by the indexing model from the issue text.