Add validation to `vn.train()` inputs

[samoliverschumacher]

This PR adds another input argument checking to vn.train(). If more than one one input is used (question and sql, documentation, ddl, plan), a validation error will be raised that notifies the user.

The Problem

• If more than one argument is sent to vn.train(), only one item will be sent to the database for later referencing
• The user wont know which one did, and since no warning is raised, won't know their mistake.

Context

• vn.train() accepts multiple kwargs for each training item, which are by default None. (question and sql, documentation, ddl, plan).
• If sql and question are not both None, or not None, a ValidationError exception is raised.
• If (sql, question) and any other of documentation, ddl, plan, are given, only one will be added, in this order: documentation, sql & question, ddl, plan (which allows for all or some).

vn = VannaDefault()
vn.train(ddl="[EmployeeId] INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
                      [LastName] NVARCHAR(20)  NOT NULL,",
         documentation="The 'employees' table contains the FirstName and LastName of employees")
# only documentation would be added

The solution

• Refactored out the current Validation, and added another check for when more than one is not None.

class VannaBase:
    ...
    @staticmethod
    def _validate_train_args(
        question: Optional[str] = None,
        sql: Optional[str] = None,
        ddl: Optional[str] = None,
        documentation: Optional[str] = None,
        plan: Optional[TrainingPlan] = None,
    ) -> None:
    # Implementation here ...
    ...
    def train(
        self,
        question: str = None,
        sql: str = None,
        ddl: str = None,
        documentation: str = None,
        plan: TrainingPlan = None,
    ) -> str:
    ...
        self._validate_train_args(question, sql, ddl, documentation, plan)

Tests;

• Followed steps as per CONTRIBUTING.md. Tests pass.