Add support for extracting spec algorithms in JSON format

[tontonialberto]

Hi,

As previously discussed with @jhnaldo during ECOOP 2025 / PLSS, I've implemented a JSON protocol for ECMA-262 algorithms extracted by ESMeta, so that users can analyze their Abstract Syntax Trees in a standardized format:

• Each algorithm is dumped in a separate file within the EXTRACT_LOG_DIR/algos_json directory
• The AST structure follows an uppercase-lowercase structure (uppercase for node names, lowercase for property names)
• Correctly serializes number literals: supports positive and negative zero, positive and negative infinity. I'm mentioning this because, by default, circe serializes Infinity values as null

The JSON protocol recursively serializes algorithms and steps. Typing information is currently represented in a stringified format - I'm working on adding support for types in JSON format.

You can serialize the algorithms in JSON format by simply executing esmeta extract -extract:log.

[jhnaldo]

Thank you for submitting this PR. I've refactored the implementation using the helper functions deriveEncoderWithType and deriveDecoderWithType. These allow us to consistently generate encoders and decoders that leverage the type name for both encoding and decoding, ensuring the encoding results are identical to the original implementation.

The only difference compared to the original implementation is in how we handle Double values, where I've reused the existing Double implementation to maintain standard behavior:

• Double.PositiveInfinity: Serializes to the String "Infinity"
• Double.NegativeInfinity: Serializes to the String "-Infinity"
• Double.NaN: Serializes to the String "NaN"
• Other Double values: Serializes as a standard Number

This keeps the serialization in line with standard JSON representations for these exceptional numeric values.
I've merged the changes, and they look great. Feel free to let me know if you have any questions.