Methods with multiple implementations such as

// File1.cs
class MultiImpl
{
    public int M1(int x) => M2(x);
    public int M2(int x) => x;
}

// File2.cs
class MultiImpl
{
    public int M1(int x) => M2(x);
    public int M2(int x) => x;
}

give rise to a single CFG, where flow splits immediately following the special entry node:

flowchart TD
1["enter M1"]
2["exit M1"]
3["exit M1 (normal)"]
4["access to parameter x"]
5["access to parameter x"]
6["call to method M2"]
7["call to method M2"]

1 --> 4
3 --> 2
4 --> 6
5 --> 7
6 --> 3
7 --> 3
1 --> 5

flowchart TD
1["enter M2"]
2["exit M2"]
3["exit M2 (normal)"]
4["access to parameter x"]
5["access to parameter x"]

1 --> 4
3 --> 2
4 --> 3
5 --> 3
1 --> 5

For data flow, this means that each of the two calls to M2 above will be able to target both bodies of M2.

This PR prevents this kind of combinatorial explosion by ranking the bodies of a method based on their file-system distance to the calling method, and only allowing flow into those bodies with the lowest rank. In order to do this, we need to track which version of a method (body) is being called, which is done by augmenting DataFlowCallables (as well as parameters) with the location of the implementation.