add ValueLattice abstraction and refactor TNum for known bits

[katrinafyi]

there is a big problem here with the Top value. the new ValueAnalysis infrastructure expects a unique value top: Value but the top in TNum requires a bit width parameter. i've arbitrarily set the bv1 TNum.top(1) as "the" top value, and this causes the current test failures.

(also, the TNum lattice is actually not really valid because values with differing widths are incompatible - meet/join will fail when given different widths.)

maybe we need an abstraction of bit-width which would allow the use of width-dependent lattices and coerce to Top when widths differ.

[katrinafyi]

i've added a TypedLattice to abstract the bitwidth-dependence of the lattices

BASIL/src/main/scala/analysis/ValueAnalysis.scala

Lines 102 to 120 in 34bed33

	/**
	* Derives a lattice for all [[ir.IRType]]s from the given type-specific [[TypedValueLattice]].
	* This is done by wrapping the "inner" lattice into [[TypedLattice.Elem]]. New
	* [[TypedLattice.Top]] and [[TypedLattice.Bot]] values are also introduced to act as
	* universal (type-independent) top and bottom elements.
	*
	* If a type-specific operation is attempted on inner lattice values of incompatible type,
	* the [[TypedValueLattice#handleConflictingTypes]] function is called.
	*
	* This [[TypedLattice]] class guarantees that the type-dependent methods of the inner
	* [[TypedValueLattice]] are only called with values of compatible type. It also guarantees that
	* the inner lattice's parameter-less [[ValueLattice#top]] and [[ValueLattice#bottom]]
	* methods are _never_ called—instead, these are implemented by returning [[TypedLattice.Top]]
	* and [[TypedLattice.Bot]], respectively.
	*/
	enum TypedLattice[T <: TypedValueLattice[T]](lattice: T) extends ValueLattice[TypedLattice[T]] {
	case Bot(lattice: T) extends TypedLattice[T](lattice)
	case Elem(inner: T) extends TypedLattice[T](inner)
	case Top(lattice: T) extends TypedLattice[T](lattice)

however, there are still some problems:

• it would be nice to allow the "inner" lattice operations to return the universal top/bot. however, this is kind of tricky because it would be mutually recursive or it would make it harder to define the cases which are width-specific or both.
• F-bounded polymorphism is really annoying because it requires subtyping for customisation. however, subclassing the TypedLattice enum causes problems especially with the enum classes (which i believe wont pick up the subtyped parent class).

[katrinafyi]

re being able to define both width-deoendent and -independent methods and being able to choose which to use, ig the thing can be the thingy blah blah method parameter dictionary of functions guh

[katrinafyi]

related to #464

[github-actions]

scalafmt check passed.

last updated at 8d66e17 (logs).