perf(*): WaitForXXX instance checks should favor loading server group…

… over cluster (spinnaker#1559)

Most WaitForXXX instance checks operate on a single server group
but pay a performance penalty by loading a cluster and filtering
to a specific server group client-side.

In particular, we noticed that clouddriver response times significant
degrade when fetching clusters containing dozens of server groups and
thousands of instances.

This PR is a targeted at `AbstractInstancesCheckTask`, there has been
no effort to improve other callers of `getCluster()` and
`getServerGroupFromCluster()` (future improvement).

fix(core): fixes issue where rolling push doesn't use existing capaci…

…ty and restores min/max/desired at the end of cycle (spinnaker#1558)

tweak(queue): Add a few more attempts when a stage fails to plan (spi…

…nnaker#1557)

Will now retry for 10 minutes with a 15s delay between attempts.

Improved logging in `TargetServerGroupResolver` in the event that a
failure is encountered.

fix(queue): Stop queuing CompleteExecution after 10 attempts (spinnak…

…er#1555)

fix(bake): root_volume_size should be ignored if null (spinnaker#1553)

- ignoring root_volume_size if null

fix(pipeline_templates): support template sources resolved from trigg…

…er context in dependent pipelines. (spinnaker#1551)

- chore(bake): Add root volume size to the request body (spinnaker#1550)

- Added a new field root_volume_size to the bake request

fix(pipeline_templates): change configuration stanza to pipelineTempl…

…ates (spinnaker#1548)

Before the configuration stanza was defined in a singular form as pipelineTemplate.
To keep it consistent with the feature name it is now changed to plural form.

Revert "fix(orca) stop just re-queueing maxed out CompleteExecution m…

…essages"

This reverts commit 6790b8c.

fix(orca) stop just re-queueing maxed out CompleteExecution messages