Incorporate extended MTK metadata for downstream use

[wsphillips]

ModelingToolkit.jl offers some additional metadata tags that we can use to more easily setup optimization. We want to be able to:

• Mark certain symbols as constants
• Mark symbols as "tunable" for inclusion in the state vector for optimization packages.
• Bounds and distributions for symbols
• Mark symbols as inputs/outputs

[wsphillips]

Basic constants added in #60

[wsphillips]

The Pinsky Rinzel network demo should demo distributions with a 10% variation on the maximal conductance of calcium channels (as in the original paper)

[wsphillips]

Distribution metadata is available in MTK but the MTK backend does not use it afaict. Therefore, we need to filter for it depending on the downstream setup (e.g. simulation, optimization, inference etc)