Add uniform series support to TimeseriesGraph for efficient multi-channel timeseries visualization

[magland]

Problem

Currently, TimeseriesGraph only supports line, marker, and interval series that require explicit time arrays. For large multi-channel timeseries data with uniform sampling (common in scientific applications), this approach is inefficient because:

1. Memory overhead: Storing redundant time arrays for each channel
2. No downsampling: Large datasets can't be efficiently visualized at different zoom levels
3. Limited multi-channel support: Each channel requires a separate line series

Solution

Add a new add_uniform_series() method to TimeseriesGraph that:

• Uniform time spacing: Only requires start time and sampling frequency (no explicit time array)
• Multi-channel support: Single series can contain multiple channels with individual colors and names
• Efficient storage: Uses downsample pyramid with min/max pairs for fast rendering at any zoom level
• Dynamic y-limits: Calculates y-axis range from visible data rather than entire dataset
• Legend integration: Each channel appears as separate legend entry

API Example

# Create multi-channel data (N timepoints × M channels)
data = np.random.randn(1000000, 4)  # 1M timepoints, 4 channels

graph = fp.TimeseriesGraph()
graph.add_uniform_series(
    name="Neural Signals",
    start_time_sec=0.0,
    sampling_frequency_hz=1000.0,
    data=data,
    channel_names=["Ch1", "Ch2", "Ch3", "Ch4"],
    colors=["red", "blue", "green", "orange"],
    width=1.0
)

[magland]

@gshvarts

This has been implemented in 0.2.13. Relates to our email conversation.

See https://flatironinstitute.github.io/figpack/basic_views_tutorial.html#uniform-timeseries

[gshvarts]

This seems like exactly what I'd want! I don't think I'll have time to try it out until tomorrow afternoon or Monday but I'll let you know when I do. the distinct colors are great too