refactor: align seed and internal naming

[tisonkun]

Summary

This PR applies the approved subset of the naming-drift audit. It establishes an explicit distinction between a full 64-bit seed and its derived 16-bit seed_hash, standardizes the set-operation constructor style, and corrects several misleading or non-idiomatic internal names.

Breaking constructor changes

Theta and Tuple set-operation constructors now follow the same convention as the other sketches:

• ThetaIntersection::default() uses the default seed; ThetaIntersection::with_seed(seed) accepts a custom seed.
• ThetaAnotB::default() uses the default seed; ThetaAnotB::with_seed(seed) accepts a custom seed.
• TupleIntersection::new(policy) uses the default seed; TupleIntersection::with_seed(policy, seed) accepts a custom seed.
• TupleAnotB::default() uses the default seed; TupleAnotB::with_seed(seed) accepts a custom seed.

The previous new(seed) and new_with_default_seed forms are removed and the migration is recorded in CHANGELOG.md.

Commit-by-commit rationale

1. 7f15f37 — S1: define seed terminology. Documents that seed is the full 64-bit hashing value, seed_hash is only the derived 16-bit compatibility fingerprint, and hash_seed must not be used as a synonym.
2. 643a358 — S2: rename the full RawHashTable seed. Changes the internal hash_seed field, parameters, and accessor to seed, removing the only third synonym while leaving Count-Min row hash_seeds intact.
3. 0646322 — S3: standardize seed-hash errors. Converts the remaining CPC and A-not-B outliers to the common incompatible seed hash wording while retaining A/B input context.
4. 1278d7b — C1: standardize set-operation constructors. Introduces consistent default-seed and custom-seed entry points across Theta and Tuple intersection and A-not-B operators; updates examples, tests, and the changelog.
5. d7db047 — P1: align CPC confidence parameter names. Uses num_std_dev on public CPC sketch and wrapper methods while retaining the mathematical kappa name inside estimator internals.
6. d073774 — P3: align ThetaUnionBuilder parameter names. Renames local parameters to factor and probability, matching the equivalent Theta and Tuple builders without changing method names or behavior.
7. 9611f37 — P4: clarify the raw A-not-B type. Renames the shared internal operator from RawAnotB to RawThetaAnotB, making its relationship to RawThetaUnion and RawThetaIntersection explicit.
8. 2a01ef3 — I2: name the hash-table threshold accurately. Replaces get_capacity with capacity_threshold, since the value controls resize or rebuild rather than reporting allocated vector capacity.
9. 01914e5 — I3: use the approved entry lookup name. Replaces get_entry with entry, following the selected Rust-style accessor name.
10. d45955c — I4: expose intersection state directly. Renames the internal is_valid field to has_result, matching the actual state-machine meaning and the existing public method.
11. b0d38fb — I5: use the plural byte-count name. Renames byte_used to bytes_used and corrects its documentation because the method returns a count.
12. b8d7242 — I6: follow mutable-accessor convention. Renames mut_surprising_value_table to surprising_value_table_mut, matching standard Rust _mut suffix usage.
13. 76a78b9 — I7: correct the raw-view documentation. Describes the view as shared by Theta-family union, intersection, and A-not-B operations rather than union alone.
14. 1264c3a — Use Default for zero-configuration operators. Replaces zero-argument new() methods with Default implementations on ThetaIntersection, ThetaAnotB, and TupleAnotB, matching the existing CPC, T-Digest, and builder conventions. TupleIntersection::new(policy) remains because its policy is required.

Validation

• cargo x test
• cargo x lint
• Targeted Theta/Tuple intersection and A-not-B tests
• Rebased onto main after chore: simplify test support code #163

Explicitly deferred

This PR does not change the RawThetaSketchView::theta or iter API, does not remove cached seed_hash fields, does not rename Coupon::from_hash, and does not include any other items from the audit that were not explicitly approved.

[tisonkun]

cc @ariesdevil in case you're working on theta/tuple sketches as well. This can be a code/logical conflict.

[ariesdevil]

Thx for informing me:)