Test optimizations for titiler-xarray

There are a number of optimizations recommend for titiler-xarray

## Software
* Titiler-xarray should use cachetools to memoize the open_dataset and/or https://github.com/developmentseed/titiler-xarray/blob/main/titiler/xarray/factory.py#L40-L45
* ​​initial consolidated metadata call is managed in the lru_cache

## Architecture
* Memory allocated to Lambda
* Lambda caching
* Caching with cloudfront

## Dataset Optimizations
* Pyramiding
* Chunk optimizations / Rechunking

Each of these should be at least one separate issue to test and estimate cost for maintaining the optimization