Episode 3, Exercise 2 proposal - Physical Units

### Brief description
Building on [the previous exercise](https://github.com/carpentries-incubator/bioimage-analysis-python/issues/19), learners now estimate the size of objects in the example image (e.g. width of nuclei).

1. Present learners with a specific Z slice from the image, and ask them to estimate the width of a few nuclei:
- first in pixel units (not sure if it's possible to interactively draw a line on matplotlib images to get a width, so may have to present the images with a a pixel grid overlay so they can manually count)
- second in physical units, using the sizes from [`bio-io`'s `.physical_pixel_sizes`](https://bioio-devs.github.io/bioio/OVERVIEW.html#metadata-reading)

2. Present learners with e.g. a YZ slice from the image, and ask them to estimate the height of a few nuclei:
- first in pixel units
- second in physical units. Ideally this dataset is anisotropic, with a larger pixel size in Z than X/Y.

### Learning objective(s)

Covers the ['Physical units from metadata' objective](https://github.com/carpentries-incubator/bioimage-analysis-python/wiki/Learning-Objectives#physical-units-from-metadata)

This exercise will demonstrate the difference between pixel units + physical units, and show learners how to extract this information with `bio-io`. If we use an anisotropic image (with a lower resolution in Z), it will also demonstrate that units can vary between different image dimensions + how to select the correct dimensions / units for our measurements.

### Volunteer(s)
@K-Meech

Review issue 19 and the linked bio-io metadata documentation first. Define the exercise using a specific Z slice and a YZ slice, with pixel- and physical-unit nucleus measurements; completion should cover dimension-specific physical sizes and, ideally, anisotropic Z resolution.

Written by the indexing model from the issue text.