repeater: Move post-ping handling into separate states

[occheung]

Bug Fix

Chained satellites setup may fail DRTIO initialization due to satellites not generating DRTIO response in time. This is caused by satellites' blocking DRTIO initialization process. This PR fixes this behavior by dividing the initialization process after ping into individual states of the FSM, which interleaves the rest of the initialization process (i.e. everything after ping) with upstream AUX handling.

Issue

After powering up all Kaslis, the following DRTIO initialization handshake may happen.

  MASTER           SAT1            SAT2
    | ----ECHO----> | ----ECHO----> |   Device pings their downstream
    |     ....      |     ....      |
    | ----ECHO----> | ----ECHO----> |   SAT2 sees ECHO
    |               | <---REPLY---- |
    | ----ECHO----> |               |
    |     ....      |               |
    | ----ECHO----> |               |   SAT1 sees ECHO
    | <---REPLY---- |               |   T+0, MASTER clears packets immediate upon receipt for 200ms (T+200ms)
    |               |     ....      |   SAT1 does the rest of the handshake with SAT2
    |               |     ....      |   T+200ms, MASTER stops clearing packets
    |               |     ....      |   SAT1 completed the handshake with SAT2
    | <---REPLY---- |     ....      |   SAT1 sends duplicate REPLY due to the queued up ECHO

Note: ECHO and REPLY stands for EchoRequest and EchoReply packets respectively.

MASTER will not expect the duplicate EchoReply packet after T+200ms, hence aborting the handshake. Link is not established between MASTER and SAT1. Satmans' blocking downstream handshake causes master to not receive timely response from satellites.

Test

Setup: RoutingTable { 0: 0; 1: 1 0; 2: 1 1 0; }, all Kasli v2.0
DRTIO links are established from master to all satellites should transceiver link ups be detected. Logs can be retrieved from satellites.

This patch should be applied to release-8 as well.