…king for new version, waiting for austraits.build update

…rough the internal data subsets for version 3.0.2 and version 4.0.0. Pivot_ needs austraits.build fixes to implenment. Some minor fix to pivot_wider for dependencies #60

Some packages needed to make plots are included in suggests. This means that core functions of package may not work.

- reuse outputs from previous function calls to reduce runtime
- reduce dataset sizes for slow functions (summarise_trait_means, trait_pivot_wider, plot_locations)
- silence some outputs

closes #62

- As documented in #79 , the Zenodo API has changed, breaking our download feature. 
- This commit updates the internals to work with the latest changes. 

Specifically: 

- the way to access json for all versions has changed (changed url structure, and for id we now use one of the record ids, rather than the conceptid)
- the call to download file has changed
- format of the API json has changed

Also

- added record id to the table of versions
- put a check in to remove "v" from any version entered by user

* changes required for v5 austraits.build

* Removed original_name for trait_pivot_wider3 for v5.0.0

* Added trait_pivot_wider for v4.x.x and code for what_version

* Making new switches for join and as_wide_table based on new versioning

* Sub switch for extract_ and recreated internal data

* Sub switches for trait_pivot_longer

* Minor fix in join_methods

* Added vars a global vars

* Removed .data calls when not needed

* Update to work with latest zenodo API (#81)


- As documented in #79 , the Zenodo API has changed, breaking our download feature. 
- This commit updates the internals to work with the latest changes. 

Specifically: 

- the way to access json for all versions has changed (changed url structure, and for id we now use one of the record ids, rather than the conceptid)
- the call to download file has changed
- format of the API json has changed

Also

- added record id to the table of versions
- put a check in to remove "v" from any version entered by user

* Recreated data so extract is passing

* Update `treatment_id` with `treatment_context_id`

* Revert "Update `treatment_id` with `treatment_context_id`"

This reverts commit 3fc6717.

* minor column name changes

Changes column names, reflecting recent changes to traits.build output.

* Update as_wide_table.R

add `any_of` to column selection within `as_wide_table` to accommodate other traits.build databases that don't have the same columns in taxon_list.csv

* Fixed getting versions and load austraits with zenodo updates and minor update with as_wide_table with removal of variable

* Fixed minor bug in get_version_latest

---------

Co-authored-by: Elizabeth Wenk <ehwenk@gmail.com>
Co-authored-by: Daniel Falster <daniel.falster@unsw.edu.au>
Co-authored-by: yangsophieee <sophie.yang917@gmail.com>

* Created lites for all main versions of AusTraits

* Passing for as_wide_table

* Passing for as_wide_table and for extract_

* expanding test suite to all 3 majors, switches for method id adjusted

* expanding test suite to all 3 majors for summarise_D

* Expanding these for all 3 versions

* Added PR trigger for dev branch

…to develop

Create a new function to check if a database is compatible with austraits 3.0 functions

In particular, this function is designed to check a database's metadata to determine if it has been built using `traits.build` versus older versions of AusTraits that were compiled by the internal `austraits.build` workflow. If a version of AusTraits (austraits.build) that is version 4.0 (or older) is detected, messages alert the user to how to download an old version of austraits.

As part of this, the helper function `util_list_to_df2` that converts nested lists to dataframes is added to austraits as an internal function. (Needed to convert specified sections of the databases' metadata into a dataframe).

Includes a test for the new `check_compatibility` function.

Closes issue #93

…d deprecating extract_dataset1

Hopefully these will work now!

I've continued to fuss with this function and there are still problems that are baffling me. I did realise that we also needed to collapse, then separate `basis_of_record` since that can vary with value_type.

Also, there are ~50 datasets where info is collapsed: these are the datasets where, for instance, there are mins and maxs for the same observation_id. For instance, `ABRS_1981` which is in the austraits5.0.0-lite, so I switched to using it. But it is huge!

Now, there is one bigger problem to discuss @dfalster. In AusTraits we now have `bin` and `range` as value types and for these, the delimiter is already `--`. So there is nothing to collapse, but the code will seek to separate these. And this will create a problem, since there won't be `--` in the other rows being separated.

I'm not sure why these functions have been passing tests until now, but it is a pretty big rethink to get them running usefully on all datasets.

Removing tests for `separate_trait_values` - somehow a large number of extra rows are being created.

This function needs a major revamp - for another day.

```
Failure (test-trait_bind_sep.R:13:3): structure of dataframes is what we expect
nrow(subset$traits) (`actual`) not equal to nrow(seperated) (`expected`).

 `actual`: 36603
`expected`: 19675
```

The functions bind_trait_values and separate_trait_values were woefully out-of-date with the traits.build structure. We're not sure how the tests were passing previously -- but they weren't any longer. Lots of bugs have been changed, but two tests are currently commented out of separate_trait_values, pending a more major reworking/scrapping of this function.

This function needs a major revamp - for another day.

```
Failure (test-trait_bind_sep.R:13:3): structure of dataframes is what we expect
nrow(subset$traits) (`actual`) not equal to nrow(seperated) (`expected`).

 `actual`: 36603
`expected`: 19675
```

…itch-deprecate