bug(import): Timestamp type mismatch when extracting paused_at_time from v3 query

[caterryan]

Summary

InfluxDB v3 returns timestamps as ISO 8601 strings, but the code expects integer nanoseconds, causing a type error during pause.

Steps to Reproduce

1. Start an import from InfluxDB v3 source
2. Pause the import while it's running
3. Observe the warning in logs

Expected Behavior

No warning. The paused_at_time should be correctly extracted from the query results regardless of timestamp format.

Actual Behavior

Warning logged, but pause still succeeds because the code falls back to current_time.isoformat().

WARN: Failed to extract paused_at_time from series: unsupported operand type(s) for /: 'str' and 'float'

[caterryan]

Additional ai generated info:

In import.py around line 1591, the code assumes max_time_ns is an integer (nanoseconds since epoch):

# Line 1591
paused_at_time = datetime.fromtimestamp(
    max_time_ns / 1e9, tz=timezone.utc
).isoformat()

However, when querying InfluxDB v3, the time column returns ISO 8601 formatted strings (e.g., "2026-02-26T15:55:06.405343+00:00"), not integer nanoseconds.

Proposed Fix

Update the timestamp extraction logic to handle both formats:

# Extract timestamp - handle both int (nanoseconds) and string (ISO 8601) formats
times = []
for row in values:
    time_val = row[time_idx]
    if time_val is not None:
        if isinstance(time_val, str):
            # ISO 8601 string - parse it
            times.append(datetime.fromisoformat(time_val.replace('Z', '+00:00')))
        elif isinstance(time_val, int):
            # Nanoseconds - convert to datetime
            times.append(datetime.fromtimestamp(time_val / 1e9, tz=timezone.utc))

if times:
    max_time = max(times)
    paused_at_time = max_time.isoformat()