Return an error for degenerate Range/ContentRange bounds instead of overflowing (#231)

[youdie006]

Fixes #231.

Problem

Range::bytes and ContentRange::bytes do unchecked u64 arithmetic on the bounds:

• Range::bytes(0u64..0u64) computes end - 1 on the exclusive end → panics in debug (attempt to subtract with overflow), and in release emits bytes=0-18446744073709551615 (a 2^64-byte span). Range::bytes(3u64..3u64) yields bytes=3-2 (start > last).
• ContentRange::bytes has the same class of bug: Excluded-start s + 1 overflows at u64::MAX, Excluded-end e - 1 underflows at 0, and an Unbounded end with complete_length: Some(0) underflows on max - 1.

Fix

Both constructors already return a Result and already Err on other invalid inputs (e.g. unbounded-start arms), so the contract-consistent fix is to return that same error for degenerate/empty/overflowing bounds using checked arithmetic, rather than panicking or emitting an invalid header.

Test

Added test_bytes_rejects_degenerate_bounds (range.rs) and bytes_rejects_degenerate_bounds (content_range.rs). Red-green verified with cargo test: before the change both panic with attempt to subtract with overflow; after, the degenerate cases return Err and valid ranges still succeed. cargo fmt --check is clean and the change adds no new clippy warnings.

---

Disclosure: I used AI assistance (Claude) while preparing this change. I reproduced the overflow, ran the tests (red-green), and take responsibility for the contribution.