[Serverless]: Provide information on how users can manage ML VCU costs

[ppf2]

Serverless Docs

Welcome to Elastic Serverless

Description

It can be helpful to add another bullet point under this section https://www.elastic.co/guide/en/serverless/current/elasticsearch-billing.html#elasticsearch-billing-managing-elasticsearch-costs that talks about the two ways to control the ML VCU costs:

• Set adaptive resources to Low to allow ML to scale down to 0 # of allocations when there are no active inference requests
• When using the inference API for Elasticsearch or ELSER, enable adaptive_allocations which will allow ML to scale down the models to 0 # of allocations when there are no active inference requests

Resources and additional context

https://www.elastic.co/guide/en/serverless/current/elasticsearch-billing.html#elasticsearch-billing-managing-elasticsearch-costs