Support creating different index types

[kvch]

This PR adds new options to create_index operation:

• method: enum for setting different index types. Default: btree. Available options: btree, hash, gist, spgist, gin, brin.
• storage_parameters: optional storage parameters for the index.

Example operation to create a new hash index with custom fillfactor:

{
  "name": "10_create_index",
  "operations": [
    {
      "create_index": {
        "name": "idx_fruits_name",
        "table": "fruits",
        "columns": [
          "name"
        ],
	"method": "hash",
	"storage_parameters": "fillfactor = 70"
      }
    }
  ]
}

Furthermore, analyze now return the index method under the name method and the definition under definition.
Please note that the definition pgroll returns is not the SQL statement pgroll runs when running create_index migrations.

{
 "idx_fruits_name": {
   "name": "idx_fruits_name",
   "unique": false,
   "columns": [
     "name"
   ],
   "method": "hash",
   "definition": "CREATE INDEX idx_fruits_name ON public.fruits USING hash (name) WITH (fillfactor='70')"
 }
}

Closes #378

[andrew-farries]

This looks good 👍

• We should update the documentation with the new index options.
• Consider adding an example migration that shows the new index options and link to it from the docs.

[andrew-farries]

We could add some new json schema tests under pkg/jsonschema to test valid and invalid index types.