FIO-10099: fix an issue where conditionally hidden layout components that are children of containers can wreak havoc

[brendanbond]

Link to Jira Ticket

https://formio.atlassian.net/browse/FIO-10099

Description

If you have a layout component as a child to a container, and that layout component contains a conditional, at some point we will ask for its path, expecting something like container.panel. Instead, because layout components don't really have a data path (instead they have absolute paths), we will get container, flagging the entire container as conditionally hidden and potentially clearing the whole thing (assuming clearOnHide is enabled). This hotfix solution addresses this by checking the component's model type - if it is none, we will use the absolute path for the component when evaluating its conditional. This PR also ensures that paths are passed to the async version of eachComponentData under certain conditions.

I do think this reveals a weakness in our treatment of conditionals - we use the "scope" object here and there to check to see if a component somewhere else has been flagged as conditionally hidden. This is brittle, and we already have an alternative mechanism by which to do so: when we determine that a component is conditionally hidden, we use its ephemeral scope property (not to be confused with the scope process results object) to label it as such. A longer term solution here would be to use that property exclusively and to no longer rely on the scope for checks of this nature.

Breaking Changes / Backwards Compatibility

n/a

Dependencies

n/a

How has this PR been tested?

Automated test

Checklist:

• I have completed the above PR template
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation (if applicable)
• My changes generate no new warnings
• My changes include tests that prove my fix is effective (or that my feature works as intended)
• New and existing unit/integration tests pass locally with my changes
• Any dependent changes have corresponding PRs that are listed above

[travist]

👍

[brendanbond]

Closing this in favor of #243