Extend eBPF kernel target with support for additional BPF helpers and more types of BPF maps

[osinstom]

This PR extends the eBPF kernel target with:

• support for defining other BPF map types
• support for BPF map-in-maps
• support for BPF maps using BPF spinlocks
• method to generate packet descriptor type (e.g. __sk_buff) depending on the eBPF hook type. It's required to support mixed deployments leveraging both TC and XDP programs.
• method to generate packet buffer's size adjustment
• BTF annotation of BPF maps, so that BPF types can be read at runtime

This PR is one from the series of PRs bringing the support for the PSA model to eBPF backend. The goal is to shrink the "main PR" with the full implementation of PSA for eBPF to facilitate review.

[osinstom]

CC: @mbudiu-vmw

[mihaibudiu]

I hope that you will soon submit a PR which tests some of these new table type too.
I am approving, but perhaps you can consider the small changes that I requested.

[osinstom]

@mbudiu-vmw I've address your review, thanks! We have ~70 test cases based on the PTF framework that verify different mechanisms of PSA impl for eBPF backend. We'll contribute them too.