ExtractSchema() does not consider option type

[psmolinsky]

Description
Schema.ExtractSchema() method does not consider the option type. This make it indistinguishable if the option is there or not.

Steps To Reproduce

1. Create a Schema with Michelson containing the option type and call the method.

const schema = new Schema({ prim: 'option', args: [{ prim: 'nat' }] });
const extracted = schema.ExtractSchema();

2. It returns:

"nat"

3. Do the same with similar schema without option and call the method

const schema = new Schema({ prim: 'nat' });
const extracted = schema.ExtractSchema();

4. It returns the same result:

"nat"

Expected behavior

The option type should be expressed in the extracted schema. For example in the step 2 the result could be "option nat" or an object { option: "nat" }. The step 4 could stay as it is.

[roxaneletourneau]

@psmolinsky
The ExtractSchema method is missing some important detail for some types and is not uniform across all tokens. (as mentioned in tickets #1303 and #1304 also)
If adding a property option for the option type, I feel that at some point it might become confusing to distinguish between the contract annotations and properties added by Taquito to the schema.
As the change mentioned in the current ticket would result in a breaking change, I thought it was a good occasion to refactor the ExtractSchema method to make it more uniform and easy to discover.
I implemented a new method named generateSchema. The intention is to deprecate ExtractSchema in favor of this new one.
For each token, generateSchema returns an object of type TokenSchema. TokenSchema has a property __michelsonType which is a string and a property schema that contains information on the schema of the subtoken when applicable.

I would like to know if the proposed solution seems convenient for you.

Here is a link to preview builds: #1316 (comment)
Some schema examples can be found in the unit tests here: https://github.com/ecadlabs/taquito/pull/1317/files

Examples:

The michelson type: { prim: 'option', args: [{ prim: 'int' }], annots: [] } will be represented as follow by the generateSchema method:

{
        __michelsonType: 'option',
        schema: {
          __michelsonType: 'int',
          schema: 'int'
        }
}

The michelson type: { prim: 'pair', args: [{ prim: 'int', annots: ['test'] }, { prim: 'string', annots: ['test2'] }], } will be represented as follow by the generateSchema method:

// Nested pair will be brought to the same level (as it is the case with the ExtractSchema)
{
        __michelsonType: 'pair',
        schema: {
          test: {
            __michelsonType: 'int',
            schema: 'int'
          },
          test2: {
            __michelsonType: 'string',
            schema: 'string'
          }
        }
}

The michelson type: { prim: 'map', args: [{ prim: 'string' }, { prim: 'int' }], annots: [] } will be represented as follow by the generateSchema method:

// schema of a map has `key` and `value` properties
{
        __michelsonType: 'map',
        schema: {
          key: {
            __michelsonType: 'string',
            schema: 'string'
          },
          value: {
            __michelsonType: 'int',
            schema: 'int'
          }
        }
}

[roxaneletourneau]

Improved the Tokenschema type and replaced the link for the preview version in the comment above.

[psmolinsky]

Great move, thanks. It makes the schema object much less ambiguous.