feat(#247): implement set operations (union, intersection, difference, symmetric difference)

[Jaskirat-s7]

Summary

Closes #247

LiteralSet lowers to a sorted constant integer array but had no visitor
for any set operation — making sets write-only constructs. This PR adds
compile-time support for all four core set operations via the existing
BinaryOp node.

What changed

src/irx/builders/llvmliteir.py

_try_set_binary_op(lhs, rhs, op_code) → bool — new helper method:

• Returns False immediately if op_code isn't |, &, -, ^
• Returns False if either operand isn't a constant integer array
• Extracts Python set[int] values from both operands
• Applies the Python set operation
• Emits a new sorted constant ir.ArrayType onto result_stack
• Promotes element type to widest integer width when operands differ

visit(BinaryOp) — one-line hook added before the existing falsy guard:

if self._try_set_binary_op(llvm_lhs, llvm_rhs, node.op_code):
    return

[Jaskirat-s7]

@yuvimittal ,i also wrote a standalone demo script that exercises all 4 set operations through the LLVMLiteIRVisitor
pipeline that constructs two LiteralSet AST nodes, feeds them through BinaryOp with each op code, and validates the resulting constant array against Python's own set algebra->
s1 = LiteralSet({LiteralInt32(1), LiteralInt32(2)})
s2 = LiteralSet({LiteralInt32(2), LiteralInt32(3)})
visitor.visit(BinaryOp(op_code="|", lhs=s1, rhs=s2))
(→ constant [3 x i32] [i32 1, i32 2, i32 3])
Demo script output — all 5 scenarios pass:

All 9 tests passing (4 existing + 5 new):

the implementation does compile-time set algebra via _try_set_binary_op where both operands are constant integer arrays, so the result is computed during IR generation with zero runtime overhead.

[xmnlab]

thanks @Jaskirat-s7

[github-actions]

🎉 This PR is included in version 1.9.0 🎉

The release is available on:

• 1.9.0
• GitHub release

Your semantic-release bot 📦🚀