Examples run slowly

[mattwthompson]

============================= slowest 10 durations =============================
455.40s call     examples/using_smirnoff_with_amber_protein_forcefield/BRD4_inhibitor_benchmark.ipynb::Cell 6
170.93s call     examples/toolkit_showcase/toolkit_showcase.ipynb::Cell 12
102.01s call     examples/toolkit_showcase/toolkit_showcase.ipynb::Cell 10
60.07s call     examples/toolkit_showcase/toolkit_showcase.ipynb::Cell 13
38.56s call     examples/forcefield_modification/forcefield_modification.ipynb::Cell 3
30.26s call     examples/conformer_energies/conformer_energies.ipynb::Cell 3
24.49s call     examples/using_smirnoff_with_amber_protein_forcefield/BRD4_inhibitor_benchmark.ipynb::Cell 3
15.98s call     examples/virtual_sites/vsite_showcase.ipynb::Cell 4
11.61s call     examples/toolkit_showcase/toolkit_showcase.ipynb::Cell 11
11.33s call     examples/toolkit_showcase/toolkit_showcase.ipynb::Cell 4

Unfortunately I can't reproduce the slowest timing on my machine. There's room for optimization in the Interchange exports, but I worry this is due to resources limitations on the runners (given that we're loading all cores with nbval and none of our code is optimized for memory usage).