Use xt.line._is_thick(element, line) to support replicas

[giadarol]

Description

The survey plot was failing on lines containing Replicas. For replicas element._isthick needs to be deduced from the parent. Reusing xt.line._is_thick(element, line_or_env), which provides isthick looking at the parent when needed.

[eltos]

Okay, but a comment considering long-term maintainability:

Wouldn't it be more useful to fix this upstream? Since isthick is already a property, it could be handled there in a transparent way? Replica could store a reference to the parent element or line (as soon as added to the line).

Another option would be to make this private line._is_thick function public, and when accessing Replica.isthick to throw an exception hinting on using line.is_thick. That would prevent similar issues from slipping through unnoticed.

[giadarol]

In general you can assume replica._parent to be there only if the replica is installed in a line and if the corresponding tracker has been built (the parents are resolved by line.build_tracker().

The function _is_thick(element, line) looks in the line to find the parent from the parent_name, also when element._parent is None.

Here we would have two options: using _is_thick(el, line) as proposed in this PR or ensuring that the line has a tracker (by building one if needed), which guarantees that all replicas have been resolved.

Both solutions would be fine with me.