rust sdk: multimodal user message content (InputContent) still missing after #2208

#2208 closes most of the event-level gap between ag-ui-core and the current spec (reasoning, activity, run outcome, resume, Tool.metadata, encryptedValue). the remaining piece on the types side is multimodal user message content: the InputContent family (text, image, audio, video, document, binary) on user messages.

I have a working port of this from downstream use, roughly 490 lines including tests. filing this first per CONTRIBUTING.md rather than opening a PR cold. I'll rebase it onto #2208's branch before opening the PR so the two do not conflict.

two questions before the PR:

1. should the full content union land at once, or would a smaller first cut (text + image) be easier to review?
2. who can be assigned, given the code owner gap in #2256?