Caret wildcard ^* is documented as matching any version but only matches 0.0.0

Hi,

I noticed a possible inconsistency in caret wildcard handling on the current master branch.

The comment above constraintCaret documents:

^* --> (any)

but ^* does not appear to match versions other than 0.0.0.

Minimal reproduction

package main

import (
    "fmt"

    semver "github.com/Masterminds/semver/v3"
)

func main() {
    c, err := semver.NewConstraint("^*")
    if err != nil {
        panic(err)
    }

    for _, raw := range []string{"0.0.0", "0.0.1", "1.0.0"} {
        v := semver.MustParse(raw)
        fmt.Printf("%s: %v\n", raw, c.Check(v))
    }
}

On the current master checkpoint
dd2b995c61c39ddd668b23ac919b04d607be35ab, I get behavior equivalent to:

0.0.0: true
0.0.1: false
1.0.0: false

I traced this far enough to see a possible explanation:

• parsing ^* rewrites the wildcard version to 0.0.0;
• dirty becomes true, while minorDirty and patchDirty remain false;
• constraintCaret then falls through to the final patch-equality check.

That seems inconsistent with the documented ^* --> (any) behavior.

Is ^* intended to mean "any version" here?

I originally found this through an independent differential conformance audit
against a small Lean 4 SemVer model. The larger audit is available here if it
is useful:

https://github.com/shumoku88-bit/semverifier/blob/main/docs/MASTERMINDS_SEMVER_CONFORMANCE_AUDIT.md

The report and analysis were prepared with substantial AI assistance, but the
minimal reproduction above is independent of the Lean model and can be checked
directly against Masterminds/semver.

Start by running the minimal Go reproduction against the current master branch, then inspect the constraintCaret implementation and the parsing path that handles wildcard versions. Confirm the intended meaning of ^* from the documented comment, and consider the issue done when behavior and regression coverage agree with that meaning.