[Ruby] Stabilize Kiota Ruby generator for production use (Open Payments SDK)

### Is your feature request related to a problem? Please describe the problem.

### Summary
Kiota's Ruby target (generation, abstractions: preview; JSON serialization,
auth: done; HTTP: preview) is close to usable but marked "not actively worked
on internally" per the Ruby tracking project. I'd like to take it from preview
to stable, motivated by generating a production Ruby SDK for the Open Payments
API (openpayments.dev, JSON only ,so FORM/MULTIPART/TEXT serialization gaps
#2077/#3032/#1049 aren't blockers for this use case).

This aligns with the Interledger Foundation's SDK Grant program, which lists
Ruby as a high-priority open track :
interledger/Grants wiki: SDK-grant-program

### Client library/SDK language

Ruby

### Describe the solution you'd like

### Proposed approach
- [ ] Generate a Ruby client from the Open Payments OpenAPI document using the
current preview generator; catalog every failure/bug encountered
- [ ] Fix generator bugs blocking correct compilation/runtime behavior
- [ ] Validate the generated client end-to-end against Open Payments' public
test environment (auth.interledger-test.dev)
- [ ] Bring generation + abstractions + HTTP from preview to stable, matching
the bar set by Go/Python/PHP

### Questions for maintainers
1. Is there a known list of blocking issues for Ruby preview→stable, beyond
the three serialization gaps already tracked?
2. Any existing branch or WIP I should be aware of before starting?
3. Open to reviewing a draft PR built against a working Ruby SDK for Open Payments?

### Additional context

_No response_

Start by generating a Ruby client from the Open Payments OpenAPI document with the current preview generator and catalog every failure or bug encountered. Use the public test environment at auth.interledger-test.dev to validate runtime behavior. Done means generation, abstractions, and HTTP are stable enough for production use, matching the stated Go, Python, and PHP standard.

Written by the indexing model from the issue text.