Hi
In a .moti template and a .motn generator, those paragraph parameters are accessed by clicking on the text directly in FCP. It's not made clear by apple, it's buried in the user manual.
Generally with text, because all those parameters, even 3D text are available in the host directly, putting a widget between the Motion parameter and the host parameter will create a conflict. There is not always good cooperation between Motion and FCP when there are common parameters like text. Another example if gradient publishing. You can publish the full gradient bar for shape layers but because FCP has gradient functions ....all the changes you make to the template gradient will be ignored when you render - if those changes add anything to what was publishing - add/remove color stoppers, change style linear/radial etc - its another case where FCP will not process changes to published parameters when rendering.
Text, generally, ...as you have seen, publishing the parameter directly is the right way. You can rig the size of text and paragraph margins internally for snapshots - FCP will not ignore those. But generally, putting a widget between the parameter and FCP and then publishing the widget will cause issues. I could be wrong about it, I'm just speaking from experience here.
For text size - what would be the benefit of rigging the size? You can group the text and rig the group scale instead, ..likewise for the text position - group and rig the group position - is the better way to go.
The sequence text behavior is the right tool if you want to rig parameters and then publish them to FCP at the same time. I recommend exploring the options you have there.
Consider that when you publish text, in the published parameter - the user can 'reset' - but in the FCP text inspector - the user can reset from there too. I recommend making sure your text position in the motion template are always at 0/0 and use the group to set layout, size, position etc. That way, your motion template text position and the fcp text position are both 0/0.