Support send and send-off methods for Agent

[pitr-ch]

like clojure has. Currently Agent supports setting an executor on initialization which globally sets the agent instance to be suitable for non-blocking only or for blocking. It would be better to be able to choose the executor when sending a function to an agent.

OneByOne could be modified to ensure sequential execution of functions over two different executors.

[jdantonio]

This sounds like a great improvement. Is this something you wanted to work on?

[pitr-ch]

I'd like to focus on migrating Actress, but if I have spare time, or if I get stuck, or after migration is done, I'll do it if nobody is quicker :) I will assign myself to the issue if I'll start working on it.

[pitr-ch]

closed by mistake

[pitr-ch]

Please review. Open questions:

• is post_off the best name?
• can we improve executor option_parsing, so when :executor options is used the executor passed in is used for both task and operation

[jdantonio]

• The method name post_off sounds odd but I can't think of a better name. One option would be post! but that wouldn't be consistent with the post/post! behavior of ActorRef. I'll give it more thought but for now I think we can use post_off so that work can continue.
• I'm comfortable with the changes you suggest to executor options parsing. That is a convenience method I added recently because I found myself duplicating code. I don't see any reason it can't evolve.

This PR broke the build for actors. Wrong number of arguments to the initializer for OneByOne. Can you please take a look at that?

Other than the broken tests I am comfortable with this PR. Please feel free to merge it.

[jdantonio]

And I should also add: Thank you very much for this update!

[pitr-ch]

• missing doc
• travis fixed

[jdantonio]

@pitr-ch Is this PR ready to merge? I promised to create a pre-release build and I'd like to include this is you feel it's ready.