Revert "ppp: Fix wrong lcp_state handling on peer reconnection"

[daniel-geotab]

Reverts #2571

This patch incorrectly mixed up the LCP state variables.

From:
https://www.eit.lth.se/ppplab/PPPdocs/PPPLCPoverview.htm

The LCP negotiation is a series of LCP packets exchanged between PPP peers to negotiate a set of options and option values when sending data. The LCP negotiation is actually two separate dialogs between two PPP peers (Peer 1 and Peer 2):

1. Peer 1 asks, negotiates, and then receives confirmation of the LCP options that are used when sending data to Peer 2. This dialog starts with Peer 1 sending a Configure-Request message and ends when Peer 2 sends a Configure-Ack message.

2. Peer 2 asks, negotiates, and then receives confirmation of the LCP options that are used when sending data to peer 1. This dialog starts with Peer 2 sending a Configure-Request message and ends when Peer 1 sends a Configure-Ack message.

The Nuttx App is peer 1 in this description. We send a CONF_REQ (in lcp_task DEBUG1(("\nSending LCP request packet - "));) and set LCP_TX_UP when we receive a CONF_ACK indicating we are ok to send because peer 2 accepted our CONF_REQ.

We also (in parallel) receive a CONF_REQ from peer 2 and set LCP_RX_UP when we send a CONF_ACK indicating we are good to receive because we accepted the request from peer 2.

The PR I am reverting incorrectly clears LCP_TX_UP when receiving a CONF_REQ claiming: "Clear LCP state to keep it in negotiation phase. LCP_TX_UP will be re-set once a CONF_ACK is received." but this is not true. We are not expecting a CONF_ACK because we are the one to be sending the CONF_ACK in this half of the exchange. It was correct to be clearing LCP_RX_UP.

[daniel-geotab]

I've attached logs of the LCP negotiations that fails when clearing LCP_TX_UP and is successful after putting back the original code.
These were captured talking to an Eagle I4915Q modem. But I have also tested on a few other quectel and wnc modems.
nuttx_ppp_good.txt
nuttx_ppp_bad.txt

[daniel-geotab]

Added missing "Signed-off-by", that I forgot the first time.