docs: clarify that partial comparator expansions are not substitution-safe

[TrueFurina]

Fixes #350 (option 2: keep the matcher behavior intentional, make the docs not
overclaim).

Problem

The Op docs describe partial comparators with "equivalent to", e.g. >I.J
equivalent to >=I.(J+1).0. At the public VersionReq::matches API these
expansions are not substitution-safe once prerelease admission is involved
across multiple comparators:

let v = Version::parse("1.3.0-alpha").unwrap();
// both true:
VersionReq::parse(">1.2, >=1.3.0-alpha").unwrap().matches(&v),
// replacing `>1.2` with its documented expansion flips it to false:
VersionReq::parse(">=1.3.0, >=1.3.0-alpha").unwrap().matches(&v),

Both reproducers from #350 were verified locally against semver 1.x
(assertions hold exactly as reported).

Fix (docs-only)

• The expansion bullets now read "expands to" instead of "equivalent to",
  describing how each form is interpreted on its own rather than claiming a
  rewrite rule.
• A short preamble on Op states explicitly that the expansions are not
  freely substitutable inside a multi-comparator VersionReq when prerelease
  admission is involved, and to prefer the form matching your intent.
• No behavior change; matcher and tests untouched.

Verification

• Both Partial comparators and their fully specified forms are not interchangeable with prerelease #350 reproducers pass as standalone tests against released semver
  (pre-change docs claim falsified, matcher behavior confirmed intentional)
• cargo test on this branch: 4 passed / 0 failed

[TrueFurina]

Friendly ping - this is still open on my side and mergeable as far as I can tell. Happy to address any remaining feedback if something is blocking it. Thanks!