There is no one solution for all, unfortunately. Here is some broader guidance.

With respect to using rigid elements for saddles

You can model at any time a rigid from the center of pipe to bottom of pipe. The only time I would recommend doing so is when radial growth of the pipe is non-negligible. It's non-negligible if the inclusion of such would significantly impact stresses in the pipe, loads on nozzles, restraint loads, spring settings, etc. In other words, if you exclude it, will it fail?

You can also use a rigid element to represent the saddle, and even include thermal growth of the saddle. The logic from before applies here.

You can also combine these two elements into one longer element of an equivalent temperature.

With respect to boundary conditions, e.g. guide, 4 way guide, anchor, etc

Generally you need the equipment drawing to render a complete answer, but I would advise that the majority of times, one anchor (or 4 way guide) and one guide will cover 95% of your situations.

In a properly designed launcher (or vessel), your sliding end should not be engaging the "limit" portion of the slot.

However, your anchor end will generally be conservative and your sliding end will be non-conservative, unless you calculate and enter restraint stiffness. As the anchor provides more give, anchor loads will mitigate, potentially allowing the slot the opportunity to actually engage, thereby reducing the anchor end's load further. Of course, the sliding end "limit" load should be fairly small, so long as we're ignoring friction effects.