Remove redundant rule for thunking #compute (#594)

We currently have two rules that `thunk` a `#compute` that is
unevaluated:

- The generic rule that `thunk`s all unfinished `Evaluation`, and
- The specific rule for `thunk`ing a `#compute`.

This removes the later.

---------

Co-authored-by: devops <[email protected]>

Author: ehildenb

kmir/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "kmir"
-version = "0.3.171"
+version = "0.3.172"
 description = ""
 requires-python = "~=3.10"
 dependencies = [

kmir/src/kmir/__init__.py

@@ -1,3 +1,3 @@
 from typing import Final
 
-VERSION: Final = '0.3.171'
+VERSION: Final = '0.3.172'

kmir/src/kmir/kdist/mir-semantics/rt/data.md

@@ -899,8 +899,6 @@ For binary operations generally, both arguments have to be read from the provide
 ```k
   syntax Evaluation ::= #compute ( BinOp, Evaluation, Evaluation, Bool) [seqstrict(2,3)]
 
-  rule <k> #compute(BOP, E1, E2, MUT) => typedValue(thunk(#compute(BOP, E1, E2, MUT)), TyUnknown, mutabilityNot) ... </k> [priority(190)]
-
   rule <k> rvalueBinaryOp(BINOP, OP1, OP2)        => #compute(BINOP, OP1, OP2, false) ... </k>
 
   rule <k> rvalueCheckedBinaryOp(BINOP, OP1, OP2) => #compute(BINOP, OP1, OP2, true)  ... </k>

kmir/uv.lock

@@ -1 +1 @@
-0.3.171
+0.3.172