Fix regression in CallClient, which caused request errors like timeouts to result in fatal errors

[cjcenizal]

Fixes #22466, a regression introduced into 6.4.0 by #20497.

[cjcenizal]

By returning the original defer.promise instead of the result of calling defer.promise.catch(), we created two parallel promise chains which would both catch the same error. By returning the result of defer.promise.catch(), we preserve the single promise chain, allowing us to prevent errors from reaching the consumer (fetch_now.js). The consumer creates fatal errors in response to rejected promises.

[elasticmachine]

💔 Build Failed

• continuous-integration/kibana-ci/pull-request

[cjcenizal]

Looks like the run:pluginFunctionalTestsRelease task failed to run.

20:55:18    │ proc [kibana]   log   [20:55:18.893] [info][status][plugin:elasticsearch@7.0.0-alpha1] Status changed from yellow to green - Ready
20:55:19    │ warn Failure loading service "remote"
20:55:19    │ERROR [POST http://localhost:9515/session / {"desiredCapabilities":{},"requiredCapabilities":{}}] unknown error: Chrome failed to start: exited abnormally
20:55:19    │        (unknown error: DevToolsActivePort file doesn't exist)
20:55:19    │        (The process started from chrome location /usr/bin/google-chrome is no longer running, so ChromeDriver is assuming that Chrome has crashed.)
20:55:19    │        (Driver info: chromedriver=2.41.578700 (2f1ed5f9343c13f73144538f15c00b370eda6706),platform=Linux 4.9.0-8-amd64 x86_64)
20:55:19    │          at Server._post (/var/lib/jenkins/workspace/elastic+kibana+pull-request+multijob-selenium/kibana/test/functional/services/remote/verbose_remote_logging.js:34:21)
20:55:19    │          at Server.createSession (/var/lib/jenkins/workspace/elastic+kibana+pull-request+multijob-selenium/kibana/node_modules/leadfoot/Server.js:362:15)
20:55:19    │          at attemptToCreateCommand (/var/lib/jenkins/workspace/elastic+kibana+pull-request+multijob-selenium/kibana/test/functional/services/remote/leadfoot_command.js:38:32)
20:55:19    │          at Promise.race (/var/lib/jenkins/workspace/elastic+kibana+pull-request+multijob-selenium/kibana/test/functional/services/remote/leadfoot_command.js:86:11)
20:55:19    │          at initLeadfootCommand (/var/lib/jenkins/workspace/elastic+kibana+pull-request+multijob-selenium/kibana/test/functional/services/remote/leadfoot_command.js:63:5)
20:55:19    │          at RemoteProvider (/var/lib/jenkins/workspace/elastic+kibana+pull-request+multijob-selenium/kibana/test/functional/services/remote/remote.js:40:29)
20:55:19    │          at <anonymous>
20:55:19    │ warn 1 processes left running, stop them with procs.stop(name): [ 'kibana' ]
20:55:19    │ info [kibana] exited with null after a minute
20:55:19 
20:55:19 Failure initializing 1 service(s)
20:55:19     at ProviderCollection.loadAll (/var/lib/jenkins/workspace/elastic+kibana+pull-request+multijob-selenium/kibana/src/functional_test_runner/lib/providers/provider_collection.js:71:13)
20:55:19     at <anonymous>
20:55:19     at process._tickCallback (internal/process/next_tick.js:188:7)
20:55:19 
20:55:19 Warning: non-zero exit code 1� Use --force to continue.

[cjcenizal]

Retest

[elasticmachine]

💚 Build Succeeded

• continuous-integration/kibana-ci/pull-request

[nreese]

Great job tracking this one down. This one has hard to pinpoint.

lgtm
code review, tested changes and verified that problem has been resolved.

[spalger]

LGTM