Add option to load schemas from file in generate command

Background

This PR will allow users to load schemas from a JSON file when running generate command, hence skipping the manual data entry. This will make it easy to move from one project to another that has shared schemas

Sample usage

kirimase generate -f <path-to-schema.json>

Notable changes

Added Zod
Added new options -f, --from-file to generate command
Moved table column types definition from src/types.d.ts to src/utils.ts so it can be reused in schema validation
The buildSchema will process array of schemas instead of a single one. The generateResources will process the schema array one by one
The printGenerateNextSteps function now process schema array. The output is updated accordingly

Input sample

Here's a sample of a valid schema.json (basically the shape of Schema)

[
  {
    "tableName": "products",
    "fields": [
      { "name": "name", "type": "varchar", "notNull": true },
      { "name": "price", "type": "number", "notNull": true }
    ],
    "index": "name",
    "belongsToUser": false,
    "includeTimestamps": true,
    "children": []
  }
]

Validation

This PR also comes with the necessary validation such as requiring the table name to be snakeCase. The validation is done by Zod. If errors are present in the schema, Kirimase will stop the execution and log the errors appropriately, such as

[
  {
    "message": "Field name must be in snake_case if more than one word.",
    "field": "fields.0.name",
    "tableName": "products.product_discounts.product_discount_analytics"
  },
  {
    "message": "Table name must be in snake_case if more than one word, and plural.",
    "field": "tableName",
    "tableName": "products.productReviews"
  }
]

Note on using Zod

Since strict mode is not enabled in this project, we will not be able to fully utilize Zod, such as type inferring

nicoalbanese / kirimase