bigquery: make InferSchema set column defaults

[lperson]

Is your feature request related to a problem? Please describe.

I'd like InferSchema to set defaults for columns.

Describe the solution you'd like

Add default= as an option for the bigquery tag. For example:

type withDefaults struct {
		Bytes         []byte         `bigquery:",nullable,default=b'hey now'"`
		Rat           *big.Rat       `bigquery:",nullable,default=3.1415"`
		RequiredBytes []byte         `bigquery:",default=b'hey now'"`
		RequiredRat   *big.Rat       `bigquery:",default=3.1415"`
		NullInt64     NullInt64      `bigquery:",default=77"`
		NullFloat64   NullFloat64    `bigquery:",default=77.24"`
		NullBool      NullBool       `bigquery:",default=false"`
		NullString    NullString     `bigquery:",default='hey now'"`
		NullJSON      NullJSON       `bigquery:",default='{}'"`
		NullTimestamp NullTimestamp  `bigquery:",default=CURRENT_TIMESTAMP()"`
		NullDate      NullDate       `bigquery:",default=CURRENT_DATE()"`
		NullTime      NullTime       `bigquery:",default=CURRENT_TIME()"`
		NullDateTime  NullDateTime   `bigquery:",default=CURRENT_DATETIME()"`
		GoTime        time.Time      `bigquery:",default=CURRENT_DATETIME()"`
		Time          civil.Time     `bigquery:",default=CURRENT_TIME()"`
		Date          civil.Date     `bigquery:",default=CURRENT_DATE()"`
		DateTime      civil.DateTime `bigquery:",default=CURRENT_DATETIME()"`
		String        string         `bigquery:",default='hey now'"`
		Bool          bool           `bigquery:",default=true"`
		Float32       float32        `bigquery:",default=3.14"`
		Float64       float64        `bigquery:",default=3.1415"`
		Int32         int32          `bigquery:",default=77"`
		Int64         int64          `bigquery:",default=7777"`
}

Describe alternatives you've considered

When creating tables

1. Modify columns in the schema returned from InferSchema
2. Execute an ALTER TABLE ALTER COLUMN statement to set the default

Both of those alternatives require extra code with the drawbacks

• Table creation wouldn't be atomic
• It would be necessary to keep extra code in sync with the table described in the structure, which is brittle

Additional context

No.