Cleanup time duration options

[karlkfi]

• Use the -Period suffix consistently. Technically, most of these should be "Delay", because they're a timer between attempts and not the duration of the event itself, but we're already using "period" for most of them. So this at least makes them all consistent without requiring rename of any external facing env vars or command flags.
• Standardize the many duplicate descriptions for duration options, at each layer of the stack.
• Add missing constants to replace hard-coded durations.
• Reafactor existing duration constants to make the names consistent.

[karlkfi]

So it turns out these test errors are actually a pretty interesting edge case... I'm just not sure why they're triggering only on this PR.

TestPublicHelm has a test where it's patching the helm config of a namespace to change the input values. But since the status doesn't have the input values and the Syncing condition doesn't either, the WaitForRepoSyncs returns immediately (within 20s), instead of waiting for the new rendered config to fully sync.

It looks like this is largely because the "commit" used by the new native helm renderer is just the name:version of the helm chart, and not a unique identifier for all inputs. We might need to hash the input values and include that in the commit string somehow.

It also occurs to me that this would happen for other spec fields as well, like the git dir. If you change the git dir after syncing has succeeded, there's no way to know if the sync status or Syncing condition are reflecting the desired state or not, because the commit hasn't changed.

[karlkfi]

The TestInlineClusterSelectorFormat failure is similar.

The validation that is failing is a Validate right after a WaitForRepoSyncs following a non-commit change. In this case, it was the cluster name that's been changed, which is stored in the config-management-system/reconciler-manager ConfigMap as the CLUSTER_NAME data field.

Even tho the reconciler-manage Pod is replaced, and the root-reconciler is updated with the new CLUSTER_NAME env var, there's no way to communicate that input data in the RootSync status. So the WaitForRepoSyncs call returns too early.

[karlkfi]

/retest

[karlkfi]

/retest

[karlkfi]

Extracted some test flake "fixes" this change uncovered:

• Wait for non-commit changes to sync #276
• Disable Syncing condition validation on error #275

[nan-yu]

/lgtm

[google-oss-prow]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: nan-yu

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [nan-yu]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment