Update iceberg_to_ducklake for DuckLake v0.3, add/improve tests

[Tishj]

Added a test where iceberg_to_ducklake is used on the entire generated catalog of tests, and (pretty much) all tables are compared against each other:

query I rowsort expected_res
select * from my_datalake.default.${table_name}
----

query I rowsort expected_res
select * from my_ducklake.default.${table_name}
----

[Tishj]

@Tmonster this is refusing to run any CI, do you know what's going on?

[Tmonster]

@Tishj maybe change the base branch to main? We can delete v1.4-andium for now

[Tishj]

We shouldn't, v1.4-andium is where v1.4.0 should be built from, and we will definitely need to keep v1.4-andium around for the LTS, main is where v1.5 development should go

[carlopi]

I would keep v1.4-andium around, we can chat tomorrow.

I think proper path is:

1. merge main: don't publish for wasm #477
2. modify this PR to point to main AND at the same time to build vs v1.4.0 (both submodule AND MainDistribution)

[carlopi]

v1.4-andium is at the moment left behind, and probably should not, but that we can fix tomorrow.

[Tmonster]

Ok, I have cancelled the last bit of running CI for now, CI here should run eventually, but potentially is clogged by all other CI on the rest of the extensions. @c-herrewijn does every extension right now have a PR up to bump submodules? I can imagine that is clogging a lot of stuff no?

[Tishj]

I think it would show the runs even if they haven't started, but this doesn't have any CI queued even

[Tmonster]

@Tishj I imagine CI passes for you locally? If so, and the builds pass, then I'm fine with merging so we can get this in quick. We can fix the tests in the next PR.
Tomorrow we can also update the v1.4 and main branches

edit: Nvm, see you just changed it. No worries then

[guillesd]

@Tishj ping me when you have this! I need to add it to the release post!

[Tishj]

Sure, my CI got canceled for some reason though?

[Tishj]

Btw this test is very slow, like, several minutes on a release build
I'm not sure if the pipeline here (or extension-ci-tools) is built to treat test_slow differently, like core does
If it doesn't, this is going to increase CI time substantially.

[Tmonster]

looks good to me