Create tables with an explicit INTEGER PRIMARY KEY.

[mithrandi]

Closes #35.

[codecov]

Codecov Report

Merging #87 into master will decrease coverage by 0.11%.
The diff coverage is 80.48%.

@@            Coverage Diff             @@
##           master      #87      +/-   ##
==========================================
- Coverage   92.48%   92.36%   -0.12%     
==========================================
  Files         106      108       +2     
  Lines       10413    10511      +98     
  Branches      838      847       +9     
==========================================
+ Hits         9630     9709      +79     
- Misses        623      640      +17     
- Partials      160      162       +2

Impacted Files	Coverage Δ
axiom/test/test_upgrading.py	99.58% <ø> (ø)	⬆️
axiom/test/historic/stub_storeID.py	0% <0%> (ø)
axiom/_schema.py	100% <100%> (ø)	⬆️
axiom/item.py	94.66% <100%> (+0.01%)	⬆️
axiom/test/test_item.py	96.31% <100%> (ø)	⬆️
axiom/test/test_xatop.py	97.13% <100%> (+0.05%)	⬆️
axiom/test/historic/test_storeID.py	100% <100%> (ø)
axiom/store.py	92.08% <88.23%> (-0.01%)	⬇️
axiom/upgrade.py	96.8% <91.66%> (-1.22%)	⬇️
axiom/_fincache.py	65% <0%> (-1.67%)	⬇️
... and 4 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 e169a79...9fdd470. Read the comment docs.

[ldanielburr]

Super nit-picky, I know, but most of the docstrings in Axiom are pretty good, so I'd like to see the usual epydoc (or sphinx or whatever is good these days) markup in here.

[mithrandi]

Good catch; it's a private method, but given that we're calling it from another "friend" module, it definitely deserves better documentation.

[mithrandi]

I pushed some documentation-type improvements.

[mithrandi]

I'm doing some testing using large existing stores that I have lying around and I've picked up some corruption issues after VACUUMing. I'm still investigating, but we definitely shouldn't merge this yet.

[mithrandi]

False alarm on the corruption, turned out to be an unrelated issue. However I would definitely still like to do more testing before landing this.

[ldanielburr]

I have one active mantissa project still running; I will try to find some time today to test it out against this PR.

[ldanielburr]

I finally got around to running this against an active mantissa project of mine; no issues to report.

[mithrandi]

After manually fixing up some schema issues (types defined in the database but not the code), I was able to do this upgrade successfully in one of my test environments and the data seems intact.

[glyph]

This is kind of amazing to see.

It looks correct to me, and I think the proof is in the pudding with respect to your vacuum testing.

Please land.