[ML] Workaround issues with second inference when IPEX is linked

[droberts195]

We have observed a couple of issues where IPEX is linked where the first inference call works but the second does not:

• IPEX v2.1.1+cpu crashes PyTorch for model where v1.13.100+cpu worked intel/intel-extension-for-pytorch#484 happens with ELSER and PyTorch 2.1
• [ML] Text embeddings produced by the Multilingual E5 base and large models are zero value elasticsearch#102541 happens with the multilingual E5 base and large models and PyTorch 1.13.1

Disabling JIT profiling avoids the problems.

[droberts195]

Important - before we merge this we need to check whether the fix of disabling JIT profiling with IPEX makes inference slower than enabling JIT profiling without IPEX. If it does then we should instead stop linking IPEX.

[droberts195]

Superseded by #2605 - we decided it's better to remove IPEX altogether.