ZMQ actors should subscribe to a single topic

[t-bast]

We use one actor per topic, but each actor previously registered to multiple topics so we received duplicate events and consumed twice the necessary bandwidth.

Thanks @btcontract for noticing this in #1789

[codecov-commenter]

Codecov Report

Merging #1793 (83f549e) into master (a8d4e07) will decrease coverage by 0.02%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #1793      +/-   ##
==========================================
- Coverage   89.37%   89.35%   -0.03%     
==========================================
  Files         143      143              
  Lines       10932    10933       +1     
  Branches      478      476       -2     
==========================================
- Hits         9771     9769       -2     
- Misses       1161     1164       +3     

Impacted Files	Coverage Δ
...ir-core/src/main/scala/fr/acinq/eclair/Setup.scala	80.23% <100.00%> (ø)
...cinq/eclair/blockchain/bitcoind/zmq/ZMQActor.scala	95.12% <100.00%> (+0.12%)	⬆️
...cala/fr/acinq/eclair/payment/relay/NodeRelay.scala	96.09% <0.00%> (-1.57%)	⬇️
...c/main/scala/fr/acinq/eclair/channel/Channel.scala	86.39% <0.00%> (-0.09%)	⬇️

[akumaigorodski]

To me it seems a bit strange that every instance of ZMQActor has logic for both rawtx and rawblock yet subscribes to a single topic, maybe it's more straightforward to use a single instance of ZMQActor which handles both topics inside?

[t-bast]

I agree that it's a bit strange to implement all topics whereas we will subscribe to a single one, but:

• There should really be two distinct actors to ensure parallelism and isolation (we really want two distinct tcp sockets)
• It's better and simpler to have all the code in the same place (instead of separating the logic in two distinct actor types)

If we opened a single socket I would totally agree with you, but given the separate sockets setup I think the existing code makes sense.