Post-landing review comments for PR 10985

[bzbarsky-apple]

Problem

I didn't get to PR 10985 before it landed. Looking at it now, I have the following questions/comments:

• StaticGroupsProvider::Decrypt should take the PacketHeader as a const reference, not by value. Implement the GroupDataProvider #11075
• StaticGroupsProvider::Decrypt should almost certainly take the PacketBufferHandle as an lvalue reference, not an rvalue reference, since it's not taking ownership of the buffer, I assume. Implement the GroupDataProvider #11075
• In SecureUnicastMessageDispatch all the MCSP stuff should go away; we should never be in a situation where !state->GetSessionMessageCounter().GetPeerMessageCounter().IsSynchronized() is true for a unicast message, and if that ever does happen we should just error out. That whole block needs to move into SecureGroupMessageDispatch somewhere.
• In SecureGroupMessageDispatch if packetHeader.IsSecureSessionControlMsg() we should probably just error out. The only control messages are MCSP messages, which are not group messages. [EDIT: This is not true. SecureGroupMessageDispatch is called for a group session, and MCSP messages do have a group session. What they do not have is a group id to filter possible decryption keys on, but we are not doing anything with group ids here yet.]

Proposed Solution

Fix those.

@jepenven-silabs

[jepenven-silabs]

Moving some of those comments to other issue