Rethrow more asio operation cancellation exception as fc::canceled_exception

[abitmore]

Rethrow more asio operation cancellation exception boost::asio::error::operation_aborted as fc::canceled_exception instead of fc::exception in read/write handlers to fix errors like below.

p2p:connect_to_task connect_to ] Failed to connect to remote endpoint x.x.x.x:x from local endpoint 0.0.0.0:x, will connect using an OS-selected endpoint: {"code":0,"name":"exception","message":"unspecified","stack":[{"context":{"level":"error","file":"asio.cpp","line":41,"method":"error_handler","hostname":"","thread_name":"fc::asio worker #1","timestamp":"x-x-xTx:x:x"},"format":"${message} ","data":{"message":"Operation canceled"}}]}

See also #249.

Note that this PR and #249 essentially revert f8472af and a part of 751777e . The commit f8472af says:

Change fc::canceled_exceptions thrown due to a socket operation being canceled into regular fc::exceptions -- we're reserving canceled_exception for canceling async tasks

But I think async task cancellation will trigger socket operation abortion, we need a fc::calceled_exception to handle it correctly.

[sonarcloud]

SonarCloud Quality Gate failed.   

0 Bugs
0 Vulnerabilities
0 Security Hotspots
0 Code Smells

0.0% Coverage
94.1% Duplication

Catch issues before they fail your Quality Gate with our IDE extension SonarLint