Support refresh interval and checkpoint location option

[dai-chen]

Description

1. Added refresh_interval and checkpoint_location option and store it in Flint metadata
2. Fixed STRING token and deleteIndex() stop job first bug
3. Extract Spark parsing to SparkSqlAstBuilder and mix-in to Flint parser

TODO

1. Didn't find easy way to verify index option in streaming job. Will add UT in next PR for index setting option.
2. Refactoring Flint metadata and build index API #24: build API refactor will come first in next PR for MV.
3. Will update user manual with new index options in next PR.

Example

spark-sql> CREATE INDEX orderkey_and_quantity
         > ON stream.lineitem_tiny (l_orderkey, l_quantity)
         > WITH (
         >   auto_refresh = true,
         >   refresh_interval = '10 seconds',
         >   checkpoint_location = 's3a://test/'
         > );

GET flint_stream_lineitem_tiny_orderkey_and_quantity_index/_mapping
{
  "flint_stream_lineitem_tiny_orderkey_and_quantity_index": {
    "mappings": {
      "_meta": {
        "name": "orderkey_and_quantity",
        "options": {
          "auto_refresh": "true",
          "refresh_interval": "10 seconds",
          "checkpoint_location": "s3a://test/"
        },
        "source": "stream.lineitem_tiny",
        "kind": "covering",
        ...
    }
  }
}

spark-sql> 23/09/21 16:06:30 WARN ProcessingTimeExecutor: Current batch is falling behind. 
The trigger interval is 10000 milliseconds, but spent 72423 milliseconds

$ aws s3 ls test
     commits/
     offsets/
     sources/

Issues Resolved

#26

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[vmmusings]

Seems like every option value is treated like a string.
Few QQs:

• If user passes some random strings in auto refresh. When will the failure occur?
• [Not related to this PR] Any thoughts on communicating syntax and other errors to user

[dai-chen]

1. Yes, it's a map of string passed to FlintSparkIndexOptions which will interpret it.
2. For the error case, currently exception will be thrown when we build DataFrame for streaming job in FlintSpark. Let me think about if Spark error is clear enough or can we validate beforehand in Add index settings option in create statement #44. Thanks for the good point!

[vmmusings]

LGTM