[issue-5494] [P-SDK] fix: reuse original trace/span IDs in project import for idempotency

[octo-patch]

Fixes #5494

Problem

When importing a project export, import_projects_from_directory generated a fresh random UUID for every trace and every span via id_helpers.generate_id(), even though the original IDs were available in the export file. This meant:

• Running the same import twice created duplicate traces (different IDs, identical content).
• Users had no way to make imports idempotent without the migration manifest, and even with the manifest a forced re-run (--force) still produced duplicates.

Solution

Pass the original id from the export file directly to client.trace() and client.span(), falling back to generate_id() only when the export contains no ID (for forward-compatibility with older export formats that predated ID storage).

# Before
trace = client.trace(
    id=id_helpers.generate_id(timestamp=original_start_time),
    ...
)

# After
trace = client.trace(
    id=original_trace_id or id_helpers.generate_id(timestamp=original_start_time),
    ...
)

The same pattern is applied to client.span(). The server handles repeat writes of the same ID gracefully (upsert or 409 dedup), so re-running import no longer creates duplicates.

The trace_id_map and span_id_map bookkeeping is unchanged — when original IDs are preserved the maps become identity maps, which is still correct for experiment recreation lookups.

Testing

• Three new unit tests added to tests/unit/cli/test_import_project.py:
  • test_trace_uses_original_id — asserts client.trace() is called with the original trace ID
  • test_span_uses_original_id — asserts client.span() is called with the original span ID
  • test_import_twice_produces_same_ids — asserts two consecutive imports use identical trace IDs
• Full CLI unit test suite (201 tests) passes without regressions.

🤖 AI-assisted contribution — code reviewed and tests verified by human author.

[baz-reviewer]

sdks/python/AGENTS.md enforces 88-char Python lines, but the new client.trace() id=original_trace_id or id_helpers.generate_id(timestamp=original_start_time), line breaks that limit—should we wrap the conditional id expression like the other keyword args?

_{Finding type: AI Coding Guidelines | Severity: 🟢 Low}

---

Want Baz to fix this for you? Activate Fixer

Other fix methods

Prompt for AI Agents:

Before applying, verify this suggestion against the current code. In
sdks/python/src/opik/cli/imports/project.py around lines 205-228 inside
import_projects_from_directory(), update the client.trace() call so the keyword argument
`id=original_trace_id or id_helpers.generate_id(timestamp=original_start_time)` does not
exceed the 88-character limit from AGENTS.md. Refactor that single long `id=` expression
by wrapping it across multiple lines (using parentheses) and keep the formatting
consistent with the other keyword arguments in the same call. Re-run lint/format checks
to confirm the line length violation is resolved.

[baz-reviewer]

The span import loop duplicates the span reconstruction flow in sdks/python/src/opik/cli/imports/experiment.py, should we refactor both to a shared helper (e.g. opik/cli/imports/utils.py)?

_{Finding type: Code Dedup and Conventions | Severity: 🟢 Low}

---

Want Baz to fix this for you? Activate Fixer

[dsblank]

@octo-patch this makes a lot of sense; thank you for the PR!

Can you either address the baz-reviewer comments, or resolve them.

Also, there are linting errors. Please fix those.