Update p2p reqresp with explicit handling of reqresp size bounds

[protolambda]

Update p2p doc to:

• Reflect ability to check much stricter SSZ size bounds (as part of the encoding strategy)
• That the parts of req/resps should be validated according to encoding strategy
• Encoding strategy describes SSZ bounds checking in more detail. Previously already checking min/max somewhat, but not enforcing it as well.
• Include FAQ item for SSZ type size bounds
• Restrict varint bytes to max 10, to avoid unlimited varint attack
• Misc. consistency, clarification and de-duplication things.

---

Related, now that we are looking more critically at the req-resp payload format/sizing:

I see we have both:

response_chunk  ::= <result> | <encoding-dependent-header> | <encoded-payload>

  +--------+--------+--------+--------+--------+--------+
  | result |   header (opt)  |     encoded_response     |
  +--------+--------+--------+--------+--------+--------+

Can we maybe remove the second one? It's nice ascii, but this and the sudden introduction of "A response therefore has the form ..." after going in-depth a lot already feels double to me. And the optional header is confusing with the earlier BNF. The header can be empty if the encoding strategy doesn't need any header data, that's all.

---

Edit: above duplicate section has been removed. The BNF-like diagram already has this exact information. And result being one byte is explained as part of the responding side.

Force push edit: bounds are now described as type size bounds instead of type bounds

[djrtwo]

looks goo. just some minor fixes

[djrtwo]

I handled the minor PR feedback (typo and markdown list formatting)