Expose an unload method for SBT nodes

[luizirber]

From #778 (comment):

Another comment: For search we only need to load internal nodes once (they will never be checked again). This helps saving total memory consumed (because we can unload the internal node after checking it). The feature/unload branch expose this in the find function, but needs more tests.

This is not so useful for gather, because internal nodes might be checked more than once, but might lead to a "low-memory" mode that always unload internal node data, or a mixed approach where we cache the internal node data for frequently accessed nodes.

A dirty version of this is in the unassigned.py I wrote for @taylorreiter, but I would rather avoid having to dig into private fields like this and have a proper method =P

This is the PR exposing the proper method.

Checklist

• Is it mergeable?
• make test Did it pass the tests?
• make coverage Is the new code covered?
• Did it change the command-line interface? Only additions are allowed
  without a major version increment. Changing file formats also requires a
  major version number increment.
• Was a spellchecker run on the source code and documentation after
  changes were made?

[codecov]

Codecov Report

Merging #784 into master will decrease coverage by 0.94%.
The diff coverage is 80.86%.

@@            Coverage Diff             @@
##           master     #784      +/-   ##
==========================================
- Coverage   79.37%   78.42%   -0.95%     
==========================================
  Files          83       82       -1     
  Lines        7035     6889     -146     
  Branches      469      469              
==========================================
- Hits         5584     5403     -181     
- Misses       1151     1185      +34     
- Partials      300      301       +1

Flag	Coverage Δ
#pytests	89.47% <86.95%> (-0.46%)	⬇️
#rusttests	48.54% <45.83%> (-0.17%)	⬇️

Impacted Files	Coverage Δ
sourmash/cli/sig/describe.py	88.67% <100%> (ø)	⬆️
sourmash/sourmash_args.py	96.03% <100%> (ø)	⬆️
sourmash/sig/__main__.py	82.5% <33.33%> (ø)	⬆️
src/core/src/sketch/minhash.rs	34.4% <45.83%> (-0.03%)	⬇️
sourmash/commands.py	83.54% <50%> (-0.15%)	⬇️
sourmash/signature.py	87.06% <88.46%> (-7.97%)	⬇️
sourmash/_compat.py	50% <0%> (-50%)	⬇️
src/core/src/index/storage.rs	69.64% <0%> (-2.58%)	⬇️
src/core/src/index/bigsi.rs	60.21% <0%> (-0.66%)	⬇️
sourmash/lca/lca_utils.py	96% <0%> (-0.62%)	⬇️
... and 9 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 4119b43...b6879ee. Read the comment docs.

[luizirber]

I defaulted it to False in Index.find(), but changed search to set unload_data to True. I think this makes sense since sourmash search only checks the index once, and if you're doing more advanced uses of search (like keeping the index in memory, and doing repeated searches) you can then set unload_data to False.

An example of this in action:

(ignore the increase in rust_sig, other things happening there)

thoughts @ctb?