Batch mode

Thanks for this project; very useful.

I would be interested in running this in batch mode; i.e. specifying the query on `stdin` or in a file. I could then make `csv-sql` part of shell pipelines and run it in continuous integration.

Something like this syntax comes to mind:

```command
$ echo '.export(results.csv) SELECT COUNT(*) from t;' | csvsql file.csv
```

Alternatively, reading the query from a file would also be useful:

```command
$ echo '.export(results.csv) SELECT COUNT(*) from t;' > query.sql
$ csvsql file.csv --load-query-from query.sql
```

If `--load-query-from` would accept `-` as a symbol for `stdin`, the line would look like this:

```command
$ echo '.export(results.csv) SELECT COUNT(*) from t;' | csvsql file.csv --load-query-from -
```

Personally, the first option (read query from `stdin` if it's not a TTY) would be my preference.

Inizia con il percorso attuale delle query interattive del comando csvsql e confronta gli esempi di stdin e --load-query-from in questa issue. Decidi quale comportamento di input supportare, incluso il caso non-TTY o il sentinel -, quindi verifica che funzioni nelle pipeline shell e nell’integrazione continua.

Scritto dal modello di indicizzazione a partire dal testo della issue.