llms/googleai: return error instead of panicking on system-only message list

[SAY-5]

Fixes #1507

When a message list contains only system messages, generateFromMessages builds an empty history (every system message is consumed into model.SystemInstruction and continued), then indexes history[n-1] with n == 0 and panics with index out of range [-1].

This adds a length check that returns a new ErrNoRequestMessage sentinel before the index, so callers get an error rather than a panic. Covered by a unit test that drives a system-only list through generateFromMessages.

PR Checklist

• Read the Contributing documentation.
• Read the Code of conduct documentation.
• Name your Pull Request title clearly, concisely, and prefixed with the name of the primarily affected package you changed according to Good commit messages (such as memory: add interfaces for X, Y or util: add whizzbang helpers).
• Check that there isn't already a PR that solves the problem the same way to avoid creating a duplicate.
• Provide a description in this PR that addresses what the PR is solving, or reference the issue that it solves (e.g. Fixes #123).
• Describes the source of new concepts.
• References existing implementations as appropriate.
• Contains test coverage for new functions.
• Passes all golangci-lint checks.

[SAY-5]

Superseded by #1509, which folds the regression test into the existing googleai_unit_test.go instead of adding a separate file. Closing this one to avoid a duplicate.