Remove GADTs and simplify module SAWCore.Term.CtxTerm

[brianhuffman]

This is a step toward removing the dependence of this module on de Bruijn indices (see #2438).

[brianhuffman]

A few high-level comments:

• This PR is intended to erase a bunch of GADT type parameters, while keeping the functional input/output behavior of all functions exactly the same (modulo changes in datatype representation).
• We introduce some new error cases in various functions. I believe these cases to all be unreachable because GHC determined that those cases were unnecessary before I erased all the GADT type index arguments.
• There are a few instances of algorithmically non-optimal operations like snoc. I expect the performance hit to be minor (this is generally code that runs once per SAWCore declaration, not in an inner loop) and temporary (see below).
• A follow-up to this PR will convert all the remaining code from explicit de Bruijn index style to use named variables instead (as in Rewrite some term-building code to avoid dealing with de Bruijn indices #2439, Adapt SAWCoreWhat4.ReturnTrip to avoid dealing with de Bruijn indices. #2443, Update SAWCentral.Builtins to avoid dealing with de Bruijn indices #2446, Adapt SAWCentral.Proof to mostly avoid dealing with de Bruijn indices #2449 and saw-core: Adapt Rewriter to avoid using de Bruijn indices in its API. #2454). It will be worth putting more effort into tidying up the coding style in that upcoming PR.

[brianhuffman]

Fun stats from git:

% git diff master | diffstat
 Module.hs       |    7
 SharedTerm.hs   |    4
 Term/CtxTerm.hs | 1156 ++++++++++++++++++---------------------------------------------
 Typechecker.hs  |   46 +-
 4 files changed, 358 insertions(+), 855 deletions(-)

So overall this PR has removed about 500 lines of code from saw-core. Woohoo!