Keep the float information in scalar_to_sql

[Sevenannn]

Which issue does this PR close?

No correlated issue

Rationale for this change

The unparser omits the float scalar value information, causing incorrect sql from unparser.

When sending a query like

select * from inventory where inv_warehouse_sk between 3.0/2.0 AND 4.0/2.0

to Datafusion, the query plan related to the scalar value is

Filter: CAST(inventory.inv_warehouse_sk AS Float64) BETWEEN Float64(3) / Float64(2) AND Float64(4) / Float64(2)`. 

However, the unparser omits the Float64() information, causing wrong query from unparser

SELECT "inventory"."inv_date_sk", "inventory"."inv_item_sk", "inventory"."inv_warehouse_sk", "inventory"."inv_quantity_on_hand" FROM "inventory" WHERE (CAST("inventory"."inv_warehouse_sk" AS DOUBLE PRECISION) BETWEEN (3 / 2) AND (4 / 2))

This will cause incorrect data in some engine, for example, in postgres 3 / 2 is 1 and 3.0/2.0 is 1.5

What changes are included in this PR?

• Add formatting step to add .0 for float values without decimal digits
• Add tests to cover the changes

Are these changes tested?

Yes

Are there any user-facing changes?

User will receive float scalar value that correctly preserves the float information.

[phillipleblanc]

Thanks @Sevenannn! I think this PR is fine to merge as is. I do have a comment about how we can keep the generated SQL a bit nicer to read for humans, while still solving the issue this PR wants to solve.

[phillipleblanc]

This is great, thanks!

[goldmedal]

Thanks @Sevenannn it looks good to me 👍

[alamb]

Thank you @Sevenannn and @phillipleblanc and @goldmedal for the reivews