feat: update DeepSeek model IDs and examples for V4 (#19398)

## Background

DeepSeek consumers need current V4 model IDs in editor autocomplete and
examples while retaining compatibility with custom and legacy string
IDs.

## Summary

Added first-class deepseek-v4-flash and deepseek-v4-pro IDs, retained
the vision model ID and arbitrary-string escape hatch, and removed
retired aliases from first-class suggestions.

## Testing

Added runtime coverage verifying both V4 IDs are forwarded unchanged and
compile-time coverage for first-class, custom, and legacy model IDs.

## End-to-end Validation

- Added and successfully ran a live example using both DeepSeek V4 Flash
and V4 Pro.

## Documentation

Updated DeepSeek provider documentation, model tables, active examples,
and historical-guide retirement notices to direct consumers to V4
models.

## Related Issues

Fixes #19378

Co-authored-by: ai-sdk-factory <308175966+ai-sdk-factory@users.noreply.github.com>
Co-authored-by: gr2m <39992+gr2m@users.noreply.github.com>