Redistricting is the process of setting the boundaries of districts. This has to be done for two principle reasons:
1. To adjust for population changes over time. This is undisputed.
2. It can be argued that this also makes sense to group together areas that have similar concerns -- in an extreme example, separating primarily agricultural districts from primarily urban districts. This is relatively uncontroversial.
Gerrymandering is a term that describes a third reason:
3. Redistricting with the intent of producing a bias towards a particular party. This is pretty uncontroversially considered bad.
It's difficult to prove that this is happening; that's what this whole case is about, or at least a subset of it. Namely, to establish the "efficiency gap" as a metric on a state-wide basis for saying whether the redistricting is trying to do #3 in the guise of doing #2.
Many proposed solutions to the redistricting problem attempt to take some basic heuristic and apply it in an objective way, but although it's possible to solve #1 pretty easily, solving #2 is inherently subjective and that's where the wiggle room for #3 is created.
If geography is the most important concern to voters, then surely voters would be able to handle grouping themselves according to geography in at-large elections with proportional representation. There could even be an explicit urban party and rural party if that is what matters most to voters.
There are residual disputes about recent Conservative changes to the criteria given to the Commission in setting boundaries, but even so this is way less egregious than the problems in the US.