WIP: Validate before dedup

[Impala123]

Description

Reorder the layers to do validate / decrypt before dedup.

Motivation and Context

With the current order of layers, a malicious node could tamper with a message which would then be discarded by the validation layer. However the dedup cache currently stores this before it is discarded by validate. Thus any un-tampered version of the same message would no longer be processed.

How Has This Been Tested?

WIP, compiles but test fails, presumably because of the mockup message being invalid

[sdbondi]

Nice - looking good. I've tested with the memorynet example (cargo run --example memorynet --release) and all looks like it's working.

There are two more items to call this done (they can go in a subsequent PR if preferred):

• A failing test for the original issue. I would look to write this as a DHT crate integration test here comms/dht/tests/dht.rs with this scenario

create valid message `m1`
`NodeA` sends `m1'` to `NodeB` where `m1'` has an invalid DHT header but with same origin mac/identical body (not even sure if this is possible - maybe clearing/changing the ephemeral_pubkey field would do it)
`NodeA` sends `m1` to `NodeB`
assert that `NodeB` has received `m1` (i.e. has not deduped it) 

• include the origin_mac field in the hash_inbound_message hash - this is so a malicious peer cannot craft an faux-encrypted message that passes validation and enters dedup cache but has the same body as the message meant for that peer, causing the peer to dedup the subsequent original message.