[any] Excessive memory usage when generating examples

Description

The json strings generated to populate the example list is done in an exhaustive manner, regardless of usage. If there are recursion or deep hierarchies of models the resulting JSON strings can get massive, leading to excessive memory usage.

With examples generated:

no examples generated:

Swagger-codegen version

Master branch at revision 955ec39bdcf469e03d00143bf57b6a53276dfe8e

Swagger declaration url

https://tripletex.no/v2/swagger.json

Command line used for generation

generate -i https://tripletex.no/v2/swagger.json
--lang java
-c ./config.json
-o ./build/generated
-Dapis -DapiTests=false -DapiDocs=false
-Dmodels -DmodelTests=false -DmodelDocs=false
-DsupportingFiles

config.json being

{
  "library": "jersey2",
  "dateLibrary": "java8",
  "java8": true
}

Steps to reproduce

Generate the client and observe the memory usage.

Related issues/PRs

• Possibly #6394

Suggest a fix/enhancement

Override DefaultCodegen#getExamples in all generatores that don't use examples at all.
Or make the example json generation be on demand instead of being generated upfront regardless of their usage.

Start at DefaultCodegen#getExamples and compare how the generators use the example list. Reproduce the issue with the Tripletex Swagger declaration and the Java jersey2 command/configuration shown above; done means example JSON is not exhaustively generated when it is unused, without breaking generators that consume examples.