PERF: Override PeriodIndex.unique

[jbrockmendel]

In trying to simplify the mess that is the PeriodIndex constructors, I found that PeriodIndex.unique is doing an unfortunate conversion to object-dtype. This PR avoids that and gets a nice speedup.

In [2]: pi = pd.period_range('1000Q1', periods=10000, freq='Q')
In [3]: %timeit pi.unique()
The slowest run took 6.25 times longer than the fastest. This could mean that an intermediate result is being cached.
1000 loops, best of 3: 226 µs per loop    <-- PR
10 loops, best of 3: 24.7 ms per loop     <-- master

[pep8speaks]

Hello @jbrockmendel! Thanks for submitting the PR.

• There are no PEP8 issues in the file asv_bench/benchmarks/period.py !

• There are no PEP8 issues in the file pandas/core/indexes/period.py !

[TomAugspurger]

Usually these merit a whatsnew entry.

[codecov]

Codecov Report

Merging #23083 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master   #23083      +/-   ##
==========================================
+ Coverage   92.19%   92.19%   +<.01%     
==========================================
  Files         169      169              
  Lines       50911    50919       +8     
==========================================
+ Hits        46939    46947       +8     
  Misses       3972     3972

Flag	Coverage Δ
#multiple	90.62% <100%> (ø)	⬆️
#single	42.3% <37.5%> (-0.01%)	⬇️

Impacted Files	Coverage Δ
pandas/core/indexes/period.py	93.44% <100%> (+0.12%)	⬆️

---

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 296c251...3bd918c. Read the comment docs.

[jreback]

thanks!