Minor refinements about software and cvterms

[ypriverol]

This PR includes the following modelling:

• How do I capture CV terms in all the parquet list structures as struct (aka record in Avro)?

@lazear in https://github.com/lazear/mz_parquet captures the CVterms for each spectrum, as Cvterm Accession and name. This approach is good but for data science I don't see people querying by CVTerms. I would love to extend that approach and use the following:

• name: Name in the ontology (required). This will enable us to search in the ontologies easily.

• value: Some of the ontology terms, for example, MS level, need a value, you can use MS level 2, Then this is optional because in the majority of the cases will be null.

• Capture the software provider at the project level.

• Some minor changes to the schema double -> float32

[lazear]

The recent version of mzparquet I've been using is in long format (each ion is it's own row) with no CV terms. Just purely a mz/intensity/isolation window/etc storage format with no/little metadata

[ypriverol]

Can you do a PR to the current dev of quantms.io with your current version of mzparquet, in that way you will be a leading contributor ;)