Ensure write_*() toml functions have the correct output format

[soniamitchell]

No description provided.

[soniamitchell]

@richardreeve if you get a chance, it would be great if you could look at this to see which bits are still accurate.

[richardreeve]

I think all of this (on the toml and hdf5 formats) is still accurate for now, but these formats may change over the next few months as we tie down exactly what data and metadata we most want to store in internal formats.

[soniamitchell]

What I wanted more clarity on was things like this:

read_estimate(data_product, component) -> float or integer
If the component is represented as a point estimate, return that value.
If the component is represented as a distribution, return the distribution mean.
If the component is represented as samples, return the sample mean.

That is, using read_estimate() to read distributions or samples.. is that something we still want?
(I haven't implemented it yet)

[richardreeve]

That seemed the most logical thing to do at the time, and we haven't discussed it since then. However, I think the things on that page most likely to change are:

1. The metadata associated with hdf5 components
2. This kind of detail about how to handle toml components
3. Whether to allow tables in toml files

I guess what I'm saying is that we may still want it, but we won't know for a month or two at least while we work through other issues.

[soniamitchell]

Cool, I'll put this on hold then