Add Python CI, update API docs, and fix Rust append behavior

[mathleur]

Description

Contributor Declaration

By opening this pull request, I affirm the following:

• All authors agree to the Contributor License Agreement.
• The code follows the project's coding standards.
• I have performed self-review and added comments where needed.
• I have added or updated tests to verify that my changes are effective and functional.
• I have run all existing tests and confirmed they pass.

[Copilot]

Pull request overview

This PR updates CI and documentation while addressing a Rust Qube::append edge case when appending into an empty receiver.

Changes:

• Add an empty-receiver fast-path to Qube::append and a regression test covering append-to-empty behavior.
• Expand Rust and Python API documentation to include additional manipulation methods (drop, squeeze, etc.).
• Update GitHub Actions CI to run Python binding tests inside an isolated uv-managed virtual environment.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
qubed/src/merge.rs	Adds an is_empty() fast-path for append by copying other into self and clearing other.
qubed/tests/test_union.rs	Adds a test for appending into an empty Qube; removes commented-out code.
docs/src/rust/qubed.md	Documents additional Rust API methods and adds examples for drop/squeeze.
docs/src/python/py_qubed.md	Adds Python docs for compress, drop, squeeze, and query methods.
.github/workflows/ci.yml	Runs Python tests in a dedicated uv venv rather than the global runner environment.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

This test locks in append producing exactly the pre-append other.to_ascii(). If append is meant to always call compress() (as described in docs / append implementation comments), the post-append ASCII can legitimately differ while remaining semantically equivalent. Consider asserting semantic equivalence (e.g., via to_datacubes()/datacube_count()), or update the expected ASCII to match the compressed form if compression is intended.

Suggested change

	let expected_ascii = other.to_ascii();
	empty.append(&mut other);
	assert_eq!(
	empty.to_ascii(),
	expected_ascii,
	"appending to an empty Qube should yield the other Qube's content"
	let expected_hash = other
	.node(other.root())
	.unwrap()
	.structural_hash();
	empty.append(&mut other);
	let empty_hash = empty
	.node(empty.root())
	.unwrap()
	.structural_hash();
	assert_eq!(
	empty_hash,
	expected_hash,
	"appending to an empty Qube should yield the other Qube's content (semantically)"

[Copilot]

The method table describes append as “merge other into self, compress, then clear other”. With the new empty-receiver fast-path in Qube::append, compress() is no longer guaranteed to run. Either keep append semantics consistent by compressing even in the fast-path, or update this documentation to describe the conditional behavior.

Suggested change

	| `append` | `fn append(&mut self, other: &mut Qube)` | Union: merge `other` into `self`, compress, then clear `other` |
	| `append` | `fn append(&mut self, other: &mut Qube)` | Union: merge `other` into `self`, then clear `other`. If `self` was non-empty, run `compress()` afterward. |

[Copilot]

This section states compress() is called automatically by append/append_many, but Qube::append now returns early when the receiver is empty without calling compress(). If that behavior is intentional, the Python docs should mention the exception; otherwise, consider ensuring append always compresses so the docs remain accurate.

Suggested change

	Compress the Qube in-place. Merges structurally identical sibling nodes, removes empty nodes, and deduplicates. Called automatically by `append` and `append_many`.
	Compress the Qube in-place. Merges structurally identical sibling nodes, removes empty nodes, and deduplicates. Typically called automatically by `append` and `append_many`, except when appending to an empty `PyQube` where `append` may return early without triggering compression. Call `compress()` explicitly if you need to ensure a canonical form.