Allow length matching with a tolerance range

[Andrew Zonenberg]

The current P&S length matching tools have a hard coded tolerance which appears to be +/- 50 um. In many cases when working with lower speed buses (e.g. HyperRAM) this tolerance is needlessly tight and results in space being wasted on meanders that could be used for signal routing.

We should add a configurable tolerance, in distance for now (but ideally in ps down the road, if/when the length matching tools become aware of propagation velocity per layer) that the tool uses to determine if traces are adequately matched.

[Andrew Zonenberg (azonenberg)]

Maybe make the tolerance a netclass property? So that you can make your RAM DQ signals be their own netclass and inherit tolerance settings from that?

[Nick Østergaard (nickoe)]

AFIK the length of the vias are not considered in the length calculations. These should probably also be handled in some way.

[Andrew Zonenberg (azonenberg)]

Via length is not considered in the current length matching implementation, however adding that should probably be a separate issue. This issue is specifically for adding tolerance to the length, unrelated to how the length is calculated.

[KiCad Janitor (kicad-janitor)]

KiCad bug tracker has moved to Gitlab. This report is now available here: https://gitlab.com/kicad/code/kicad/-/issues/2280