Execution polling intervals not set correctly

[dpmatthews]

The execution polling intervals are not set correctly when you specify a list of intervals in combination with an execution time limit.

Example suite:

[scheduling]
    [[dependencies]]
        graph = root
[runtime]
    [[poll]]
        [[[job]]]
            execution polling intervals = PT30S, PT1H
    [[limit]]
        [[[job]]]
            execution time limit = PT30M
    [[limit_poll]]
        [[[job]]]
            execution time limit = PT30M
            execution polling intervals = PT30S, PT1H

Relevant suite log entries:

[poll.1] -health check settings: execution timeout=None, polling intervals=PT30S,PT1H,...

This looks fine.

[limit.1] -health check settings: execution timeout=PT40M, polling intervals=PT31M,PT2M,PT7M,...

This makes sense (although I'm not sure why the execution timeout is set - don't think this is documented anywhere).

[limit_poll.1] -health check settings: execution timeout=PT40M, polling intervals=60*PT30S,PT1M,PT2M,PT7M,...

This is wrong - it looks like only the first item in the execution polling intervals list is being used.

Tested with 7.8.6 & repeated on master.

[hjoliver]

Update: this was giving the right answer for the wrong reasons on current master, but will revert to giving the wrong answer for the right reasons once #3981 is merged.

[dpmatthews]

Bug fixed in #4844.
Execution timeout issue moved to #4859.