DGB node update messed up miner on HMpool

[devinestrada-max]

This is from someone on the hmpool team:

I am not sure but I can see what's going on. I had to kick your friend so he re connects to watch how his handshake process happens. His miner and yours is requesting a difficulty far below the norm and far below what our pool would accept. Your miners are requesting 0.00015 when the default difficulty is 0.01. My suggestion would be to tell Nerdminer about this. If by chance they respond asking you to use the d=0.00015 as the password to set the difficulty manually, please tell them that this will not work because our pool will automatically set the difficulty to 0.01.

[Gheop]

The pool operator is describing a real bug, and it is narrower than "the firmware
requests a low difficulty". The firmware requests the right difficulty once, then
spends the rest of the connection asking for the wrong one.

Two places send mining.suggest_difficulty. Connection setup gets it right,
src/mining.cpp:347:

tx_suggest_difficulty(client, currentPoolDifficulty);

The socket keepalive does not, src/mining.cpp:173:

Serial.println("  Sending  : KeepAlive suggest_difficulty");
tx_suggest_difficulty(client, DEFAULT_DIFFICULTY);   // 0.00015, hardcoded

That runs every KEEPALIVE_TIME_ms, which is 30 seconds. So the pool sends
mining.set_difficulty, the firmware stores it in currentPoolDifficulty and mines
against it correctly, and then goes on asking for 0.00015 twice a minute regardless,
forever.

I can see it on my own miners. They run on a pool that imposes a difficulty of
100000, and the serial log still shows this every 30 seconds:

  Sending  : KeepAlive suggest_difficulty
  Sending  : {"id":7,"method":"mining.suggest_difficulty","params":[0.00015]}

Two rounds later the id is 8, then 9. It never stops and it never reflects what the
pool told us.

From the pool side that is indistinguishable from a client ignoring
set_difficulty, which is exactly how it was read here. The miner is not actually
misbehaving on the mining side, it just keeps saying something it does not mean.

Fix

Pass the current difficulty through to the keepalive instead of the compile-time
default:

https://github.com/Gheop/NerdMiner_v2/tree/fix/keepalive-difficulty-805

Three lines: one parameter added to checkPoolInactivity(), the hardcoded constant
replaced, the call site updated. Builds for NerdminerV2 and TTGO-T-Display.

Worth saying that using mining.suggest_difficulty as a keepalive at all is
questionable. The comment above the code is honest about it: "send something to pool
to hold socket opened". A suggestion the client repeats forever is not a ping, and
some pools may reasonably rate-limit or penalise it. But repeating the negotiated
value is at least truthful, and it is a three-line change rather than a protocol
redesign. If HMpool would rather see nothing at all on that timer, that is a
one-line follow-up.