docs(logs): clarify attribute value types for Python logs (#17921)

## Summary

Clarifies which value types are supported for Python log attributes,
after a user reported that array-valued attributes (e.g. `list[int]`
like `workflow_ids`/`detector_ids`) silently show up empty in the Logs
UI.

## Findings (validated against a real project)

Sentry's log storage (EAP) only retains **scalar** attribute types —
string, number, boolean. The Python SDK
(`sentry_sdk.utils.format_attribute`) sends a homogeneous scalar
`list`/`tuple` as a real **array**, which the backend does not store, so
it appears empty. Mixed lists and `dict`s are `repr`'d to strings and
*do* survive. (For contrast, `@sentry/core` stringifies all
non-primitives via `serializeAttributes(attrs, true)`, so JS users see
arrays as JSON strings — the two SDKs diverge here.)

## Change

Adds an `<Alert>` to the Python logs usage include documenting that
attribute values should be primitives, explaining the array behavior,
and recommending serializing complex values (e.g. `json.dumps(...)`)
before logging.

A separate `sentry-python` issue tracks the SDK-level inconsistency.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Co-authored-by: Kyle a.k.a. TechSquidTV <git@techsquidtv.com>
Co-authored-by: Shannon Anahata <shannon.anahata@gmail.com>

Author: 4 people

platform-includes/logs/best-practices/python.mdx

@@ -69,7 +69,8 @@ sentry_sdk.logger.info(
 
         # Business context
         "order_value": 149.99,
-        "feature_flags": ["new-checkout", "discount-v2"],
+        # Join lists into a string — array values aren't stored as attributes
+        "feature_flags": ",".join(["new-checkout", "discount-v2"]),
     }
 )
 ```

platform-includes/logs/usage/python.mdx

@@ -30,3 +30,9 @@ sentry_sdk.logger.error(
     }
 )
 ```
+
+<Alert title="Attribute value types">
+
+Attribute values should be primitives — strings, numbers, or booleans. Sentry's log storage only retains these scalar types, so other values are handled inconsistently. A `list` or `tuple` of same-typed scalars (for example `[1, 2, 3]`) is sent as an array, which is **not currently stored** and shows up empty in the Logs UI. Mixed lists and `dict`s are stored as their string representation. To keep a complex value searchable, serialize it to a string before logging, for example `json.dumps(workflow_ids)`.
+
+</Alert>