pytests: fix flaky checkpoint sync test

In sync tests, wait until the most recently checkpoint, the one with all
the data is synced.
This didn't show up previously, as all the data was inserted and
executed within the checkpoint interval of 5s, and this checkpoint was
synced within the interval of 10s.

Signed-off-by: Abhinav Gyawali <[email protected]>

[python] In tests, print deployment status while waiting for it to ch…

…ange.

We've seen timeouts in places where they shouldn't happen.  This ought to
help understand what's happening.

Signed-off-by: Ben Pfaff <[email protected]>

[adapters] NATS: Add more error context for create consumer

Signed-off-by: kfollesdal <[email protected]>

[ci] apply automatic fixes

Signed-off-by: feldera-bot <[email protected]>

py: add illegal argument tests for cast function

Signed-off-by: rivudhk <[email protected]>

[adapters] Panic in logging code.

A user reported a panic with the following call stack:

```
thread 'circuit-thread' (17) panicked at /home/ubuntu/.cargo/registry/src/index.crates.io-1949cf8c6b5b557f/itertools-0.14.0/src/format.rs:95:21:
Format: was already formatted once
stack backtrace:
   0: std::panicking::begin_panic
   1: <itertools::format::Format<I> as core::fmt::Display>::fmt
   2: core::fmt::write
   3: core::fmt::write
   4: alloc::fmt::format::format_inner
   5: <sentry_tracing::converters::FieldVisitor as tracing_core::field::Visit>::record_debug
   6: tracing_core::field::ValueSet::record
   7: sentry_tracing::converters::extract_event_data
   8: sentry_tracing::converters::extract_event_data_with_context
   9: sentry_tracing::converters::breadcrumb_from_event
  10: <sentry_tracing::layer::SentryLayer<S> as tracing_subscriber::layer::Layer<S>>::on_event
  11: tracing_core::event::Event::dispatch
  12: dbsp_adapters::util::LongOperationWarning::check
  13: dbsp_adapters::controller::CircuitThread::run
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
```

which points to this line:

```rust
            self.checkpoint_delay_warning
                .get_or_insert_with(|| LongOperationWarning::new(Duration::from_secs(1)))
                .check(|elapsed| {
                    info!(
                        "checkpoint delayed {} seconds because of: {}",
                        elapsed.as_secs(),
                        reasons.iter().format(", ")
                    )
                });

```

The panic can happen if the iterator returned by the `format` method is
evaluated twice. This happens inside the internals of the tracing crate. Not
sure why we haven't seen this in the past. The fix just converts the iterator
to string, so there is no way it can get evaluated twice.

Signed-off-by: Leonid Ryzhyk <[email protected]>

[datagen] fix binary generation

we would generate i8 in this case but our code expects
u8.

Signed-off-by: Gerd Zellweger <[email protected]>

[adapters] Fix data interleaving in HTTP ingress connector.

Thanks to Bruno Rucy @brurucy and Abhinav Gyawali @abhizer for help with
this issue.

Fixes: #3495
Signed-off-by: Ben Pfaff <[email protected]>

[SQL] Fix several compiler bugs discovered while testing more complex…

… programs

Signed-off-by: Mihai Budiu <[email protected]>

[adapters] Add `verbose` config setting for delta.

The new `verbose` flag can be used to log detailed progress at the INFO level
for a specific connector.

Currently we only log add/delete actions in follow and cdc modes.

Signed-off-by: Leonid Ryzhyk <[email protected]>