Only validate name, not all components

[idubinskiy]

Despite the spec stating that each component is a name, the reference Java implementation only validates the actual name field. Match the Java implementation to be more compatible with OCF files written by other languages.

Specifically this has bitten us with a namespace field written by Spark that starts with a leading dot.

[jung-kim]

FYI, library that writes this dot prefixed namespace name is the https://github.com/databricks/spark-avro.

I'm not sure what is the what is right or wrong because reading the documentation seemingly this library is actually following the schema specification and other have relaxed enforcements of naming convention.

Below is the dataframe schema that caused dot prefixed namespace name along with the avro schema generated by the spark-avro.

root
 |-- usageData: array (nullable = true)
 |    |-- element: struct (containsNull = true)
 |    |    |-- app: string (nullable = true)

{
  "type": "record",
  "name": "topLevelRecord",
  "fields": [
    {
      "name": "usageData",
      "type": [
        {
          "type": "array",
          "items": [
            {
              "type": "record",
              "name": "usageData",
              "namespace": ".usageData",
              "fields": [
                {
                  "name": "app",
                  "type": [
                    "string",
                    "null"
                  ]
                }
              ]
            },
            "null"
          ]
        },
        "null"
      ]
    }
  ]
}

[karrick]

I'm actually going to pull this in, and then put a feature flag on it.

I think while it's imperative to stick to the schema definition, we would be creating a useless library if it did not properly work with schemas created by other programs.

[karrick]

Release v2.1.0 has the exported variable RelaxedNameValidation, which when true, allows the first component of a schema name to be the empty string. The default value for this variable is false, and your program ought to set it before building any codecs when this behavior is desired.