fix(isthmus): handle Subqueries/set predicates with field references outside of the subquery

[mbwhite]

This is resolving issue #382 found with TPC-H 17, when converting back to SQL from Substrait
The subquery references fields in an outer scope, the calcite correlation variables where referenced by Subsrtrait's 'outer reference'

However when converting back to calcite from a plan with these 'outer references' they were ignored.

Notes::

• I've added two new TPC-x tests, these are similar the existing tests, but do conversions SQL to Substrait and Substrait to SQL
• Additional SQL using subqueries has been added; would like to add some more. Suggestions welcome.
• Added these as 1xx.sql files but could move to a separate folder?

[vbarua]

Left some questions

[mbwhite]

@vbarua questions have been answered; I've updated the test cases to better separate concerns.

found the hidden setting in vscode to change the formatted back to spotless!

[vbarua]

minor: avoid printing during test. I suggest assertNotNull instead

[mbwhite]

also changed the other existing tests where this occurred.

[vbarua]

What is this for? It's never used.

[vbarua]

Do we need to track or handle the fact the there might be multiple Filters that can add correlation variables? We only ever add to this list.

How would we know which ones came from which input?

This is one the issue I had in mind when I mentioned
https://github.com/substrait-io/substrait-java/pull/383/files#r2038429378

[mbwhite]

Honestly the Calcite docs don't really help with the scope of these variables. There is some concept of 'namespace' eg from this error .

All correlation variables should resolve to the same namespace. Prev ns=org.apache.calcite.sql.validate.IdentifierNamespace@d36c1c3, new ns=org.apache.calcite.sql.validate.IdentifierNamespace@96abc7

which came from

select
    c1.c_name,
    o1.o_orderstatus,
    o1.o_totalprice
from
    customer c1,
    orders o1
where
    o1.o_custkey = c1.c_custkey
    and o1.o_totalprice > (
        select
            avg(o_totalprice)
        from
            orders o2, customer c2
        where
            o2.o_totalprice < c1.c_acctbal
            and o2.o_totalprice > (
                select
                    avg(c3.c_acctbal)
                from
                    customer c3
                where
                    c3.c_custkey = o2.o_custkey
                    and c3.c_address = o1.o_clerk
            )
    );

change the last line to c3.c_address = o2.o_clerk and it's ok..

[mbwhite]

Implications is that each relation and it's immediate subexpression is the same namespace.