[3.13] AI Gateway: Document support for a basic cross-LLM tokenizer for token rate-limiting

## Description

Add documentation for a basic tokenizer option that approximates token counts across different LLMs, enabling pre-request token estimation in the token rate-limiting plugin. This is to help users predict and enforce token budgets before sending a request to the model, without depending on provider-specific tokenizers.

## Definition of Done

[TBC]

* Update token rate-limiting plugin docs to describe the basic tokenizer option
* Provide guidance on expected accuracy and use cases
* Add configuration example showing how to enable it
* Add brief how-to for estimating and enforcing token limits before request execution

## Information

Related ticket: https://konghq.atlassian.net/browse/AG-527

### Size

S

Start with the token rate-limiting plugin documentation and review the existing configuration examples and tokenizer options. Document the basic cross-LLM tokenizer, its expected accuracy and use cases, and how to estimate and enforce limits before requests execute. Confirm the final scope against related ticket AG-527, since the Definition of Done is still marked TBC.

Written by the indexing model from the issue text.