[LLDB] silently allows a malformed DWARF expression, uses DW_OP_reg for DWARF expressions

Per the manual (https://dwarfstd.org/doc/DWARF5.pdf):

DW_OP_reg can only be used in a simple location description, not in a dwarf expression. However, LLDB allows it in a dwarf expression:

https://github.com/llvm/llvm-project/blob/067b6d90b64e8f43c9a2388530b7f98911d77421/lldb/source/Expression/DWARFExpression.cpp#L1959-L1967

It should show an appropriate error when it encounters such malformed dwarf.

### Example

```cpp
#include
#include

void thrower() {
throw 1lu;
}

void middle() {
__asm__(".cfi_escape 0x16, 0x03, 0x03, 0x53, 0x32, 0x24");
thrower();
}

int main() {
try {
middle();
} catch (uint64_t i) {
uint64_t rbx;
__asm__("mov %%rbx, %0" : "=r"(rbx));
std::cout << "RBX: " << rbx << std::endl;
}
}
```
Compile with:
```bash
clang++ -g -o main main.cpp
```

Run:
```
$ lldb main --batch -o "b thrower" -o "run" -o "up" -o "up" -o 'expr $rbx'
(lldb) expr $rbx
(unsigned long) $0 = 0
```

Comparatively, GDB shows the error:
```
$ gdb main --batch -nx -ex "b thrower" -ex "run" -ex "up" -ex "up" -ex 'p $rbx'
DWARF-2 expression error: `DW_OP_reg' operations must be used either alone or in conjunction with DW_OP_piece or DW_OP_bit_piece.
```

Start with the linked handling in lldb/source/Expression/DWARFExpression.cpp at lines 1959-1967, then reproduce the behavior using the supplied C++ example and LLDB batch commands. Done means malformed DWARF using DW_OP_reg in an expression is rejected with an appropriate error instead of evaluating as zero.

Written by the indexing model from the issue text.