database catalog and semantic model interoperability

[simonpcouch]

Related to #55. Just to actually see what one e2e take on this could feel like.

Treat a connection to Snowflake/Databricks plus a schema path as minimally sufficient:

db <- DBI::dbConnect(odbc::snowflake())

source <- data_source(
  db,
  options = data_source_options(
    include = DBI::Id(...)
  )
)

agent <- commons(chat_anthropic(), source)

commons then discovers the schema and native semantic models without requiring a data dictionary.

[simonpcouch]

Have been testing with:

db <- DBI::dbConnect(odbc::snowflake())

DBI::dbExecute(
  db,
  'USE WAREHOUSE "MC_TUTORIAL_WAREHOUSE"'
)

source <- data_source(
  db,
  options = data_source_options(
    include = DBI::Id(
      catalog = "LENDING_CLUB_CLONE",
      schema = "PUBLIC",
      table = "DATABOT_TEST"
    )
  )
)

ui <- page_fillable(
  commons_ui("chat")
)

server <- function(input, output, session) {
  agent <- commons(
    chat_anthropic(), 
    source
  )

  commons_server("chat", agent)
}

shinyApp(ui, server)