x86 inline assembly should provide an alternative for Intel syntax

Instead of https://github.com/boostorg/interprocess/blob/a0c5a8ff176434c9024d4540ce092a2eebb8c5c3/include/boost/interprocess/detail/atomic.hpp#L174
we should have
```c
"xadd { %1, %0 | %0, %1 }":
```

Open include/boost/interprocess/detail/atomic.hpp at line 174 and inspect the existing x86 inline-assembly operand template. Update that template to provide the Intel-syntax alternative shown in the issue; the work is done when the assembly supports both operand-order syntaxes without changing the atomic operation.

Vom Indexierungsmodell aus dem Issue-Text verfasst.