Dependency reduction algorithm for multi-variable statement

So that we actually can solve arbitrary `statement` for an arbitrary set of variables. That's a very draft version of what I mean (for three variables):
In other words, first we solve the statement over the first variable, getting `x = f1(y, z, ...)`. Then over the second one, getting `y = f1(z, x, ...)`, etc. Every variable depends on `n - 1` of others. Then we start our `DRA` and reduce the number of arguments in each by one. Finally we will end up having `x = k1(x)`, `y = k2(y)`, ...

The issue proposes a dependency reduction algorithm (DRA) for solving statements over multiple variables, but it names no source files, entry points, or tests. Start by locating the existing statement-solving and DRA implementations, then clarify the algorithm with concrete multi-variable examples. Done should include an agreed implementation scope and tests showing arbitrary-variable solving and dependency reduction.

Written by the indexing model from the issue text.