fix(eap): Ensure reliabilities are returned more consistently

[Zylphrex]

When extrapolating, if there's no data, we skip returning reliabilities. This is problematic as we want to line up each reliability value with the corresponding result so those arrays must be the same length. This fills it with RELIABILITY_UNSPECIFIED.

[Zylphrex]

I didn't want to change too much existing behaviour in this PR but for whatever reason, when using a non aggregate in a timeseries query, we always consider as data_present=True.

[kylemumma]

I saw this come up in discuss-sns when a user was confused about it. Seems like it might be a bug.

[Zylphrex]

Yeah, I didn't want to change too many behaviours in this one PR so I kept this behaviour for now but it does feel like a bug.

[onkar]

@kylemumma can you look into this and file a ticket on eap-planning to address this? I am not following what the bug is really but if either of you have an idea of the issue, we should track it and even fix it. Let's not leave this unaddressed.

[codecov]

❌ 1 Tests Failed:

Tests completed	Failed	Passed	Skipped
2784	1	2783	5

View the top 1 failed tests by shortest run time

tests.web.rpc.v1.test_endpoint_trace_item_table.test_endpoint_trace_item_table_extrapolation.TestTraceItemTableWithExtrapolation::test_count_reliability_with_group_by

Stack Traces | 0.31s run time

Traceback (most recent call last):
  File ".../v1/test_endpoint_trace_item_table/test_endpoint_trace_item_table_extrapolation.py", line 378, in test_count_reliability_with_group_by
    response = EndpointTraceItemTable().execute(message)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File ".../web/rpc/__init__.py", line 83, in execute
    return self.__after_execute(in_msg, out, error)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File ".../web/rpc/__init__.py", line 104, in __after_execute
    raise error
  File ".../web/rpc/__init__.py", line 79, in execute
    out = self._execute(in_msg)
          ^^^^^^^^^^^^^^^^^^^^^
  File ".../rpc/v1/endpoint_trace_item_table.py", line 320, in _execute
    column_values = _convert_results(in_msg, res.result.get("data", []))
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File ".../rpc/v1/endpoint_trace_item_table.py", line 215, in _convert_results
    extrapolation_context.reliability
  File ".../local/lib/python3.11/functools.py", line 1001, in __get__
    val = self.func(instance)
          ^^^^^^^^^^^^^^^^^^^
  File ".../rpc/common/aggregation.py", line 189, in reliability
    lower_bound, upper_bound = _calculate_approximate_ci_percentile_levels(
                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File ".../rpc/common/aggregation.py", line 607, in _calculate_approximate_ci_percentile_levels
    return (lower_index / n, upper_index / n)
            ~~~~~~~~~~~~^~~
ZeroDivisionError: float division by zero

To view more test analytics, go to the Test Analytics Dashboard
📢 Thoughts on this report? Let us know!

[kylemumma]

I liked your use of comments and your code was easy to read

[kylemumma]

I saw this come up in discuss-sns when a user was confused about it. Seems like it might be a bug.