Better way to support various schema types

[cybertk]

In current design, ramlev detects schema type by scanning the content of schema, it's not good and accurate if ramlev is used to support any kind of schema.

Proposal,

• Make ramelv un-aware of specific schema type, let specific schema validator pluggable.
• Provide a CLI options --schema-type to specify schema type, which will used to select the underling schema validator, and delegate the validating job to pluggable schema validator