read_from_socket should always retry if it returns with errno=EAGAIN

[erlingrj]

Currently, it will only retry a certain number of times before deeming the socket connection for faulty. This will then close this socket, and in the case of the RTI, it will then deem a federate disconnected.

I have observed this happening with federates that don't produce any signals/messages for a long time. E.g. the ones that have only intermittent events due to a physical action.

I will address this issue soon in a PR, logging it here just-in-case.

[erlingrj]

Actually, bounding the number of retries was done in this PR #323, @edwardalee do you remember the reasoning?

What I have seen is that when there is that the call to read will return with EAGAIN after a while, so if there is periods with no data arriving from a federate it will just close the connection to it and drop any data destined for it.

[edwardalee]

I agree that EAGAIN should always result in a retry.