Using thicker substrate to get higher Er tolerances and other similar methods.

1. Using thicker substrate
good: wider bandwidth for patch antenna / band-pass filter (higher Er tolerance)
bad: radiation losses

2. Using thinner lambda/4 stubs
good: wider bandstop
bad: can't use 50Ohm lines?

3. Using SIW instead of 90deg branch-line coupler
good: simple matching?
bad: ...?

The eps_r tolerances don't change with substrate thickness. What changes is the error from etching/manucfacturing tolerances, because thick substrate requires wider lines (for same impedance) and then the relative error from width tolerances (etching precision) is smaller.

Regarding SIW, there was an interesting overview paper on substrate intregrated waveguide filters in IEEE microwave magazine Vol 15 No 6 (September 2014). Most examples where in the range 15GHz up to 60GHz, but miniaturized filters an alumina ceramic where also discussed.