GCS incr source does not handle pubsub message properly

Gcs event source uses schema converter from spark and won't handle field name with hyphen in nested column. a sample message

{code:java}
23/04/03 19:23:45 DEBUG GcsEventsSource: msg: {
"kind": "storage#object",
"id": "",
"selfLink": "",
"name": "",
"bucket": "",
"generation": "1680505551370137",
"metageneration": "1",
"contentType": "application/octet-stream",
"timeCreated": "2023-04-03T07:05:51.373Z",
"updated": "2023-04-03T07:05:51.373Z",
"storageClass": "STANDARD",
"timeStorageClassUpdated": "2023-04-03T07:05:51.373Z",
"size": "6707",
"md5Hash": "",
"mediaLink": "",
"metadata": {
"goog-reserved-file-mtime": "1680503048"
},
"crc32c": "",
"etag": ""
}
{code}

and it throws

{code}
Exception in thread "main" org.apache.avro.SchemaParseException: Illegal character in: goog-reserved-file-mtime
at org.apache.avro.Schema.validateName(Schema.java:1571)
at org.apache.avro.Schema.access$400(Schema.java:92)
at org.apache.avro.Schema$Field.(Schema.java:549)
at org.apache.avro.SchemaBuilder$FieldBuilder.completeField(SchemaBuilder.java:2258)
at org.apache.avro.SchemaBuilder$FieldBuilder.completeField(SchemaBuilder.java:2254)
at org.apache.avro.SchemaBuilder$FieldBuilder.access$5100(SchemaBuilder.java:2150)
at org.apache.avro.SchemaBuilder$GenericDefault.noDefault(SchemaBuilder.java:2557)
at org.apache.hudi.org.apache.spark.sql.avro.SchemaConverters$.$anonfun$toAvroType$2(SchemaConverters.scala:205)
{code}

This is a problem with org.apache.spark.sql.avro.SchemaConverters#toAvroType

## JIRA info

- Link: https://issues.apache.org/jira/browse/HUDI-6028
- Type: Bug

Start at the GcsEventsSource entry point and trace how the Pub/Sub message reaches Spark's org.apache.spark.sql.avro.SchemaConverters#toAvroType. Reproduce the sample message with the hyphenated nested metadata field; done means the GCS incremental source processes it without the reported SchemaParseException.

Written by the indexing model from the issue text.