outer_boundary_max_epsdis and ah_max_epsdis, when set to negative values, are supposed to turn the respective maximum off. Both have a negative default value (but also need another parameter to enable dissipation in the respective case). Both don't include a check for negative numbers, and do set the dissipation strength to -1 in those cases.
This wasn't observed up until recently due to an incorrect scheduling of the AH routines (which both testsuites use), which masked the output as if no extra dissipation was applied.
The fix is to include a check for the maximum being negative, and turning this 'capping' off. This does not only change the output of epsdis itself, as epsdis is now set differently, which causes substantial differences inside the horizon and at the outer boundary.
This fix, however, would produce too large values for dissipation with the current parameter files, and the affected parameters are really expected to be set when using these location-dependent dissipation variants. Thus, the testsuites have been adjusted to actually do set a maximum in both cases, now producing different results.
Both changes are being prepared for a push.