Be more intelligent about caching the number of benefits records

[hancush]

In its original implementation, the search view performed a count of the entire benefits table every time a search was issued. This was a real drag on performance. The total number of benefit records is now hard-coded into the search view to eliminate the duplicate query.

bga-pensions/pensions/views.py

Lines 428 to 429 in ed99d1c

	# TODO: Cache this, rather than hardcoding
	total_records = 3713872

We should be smarter about this. Perform and cache the count when data is imported, and retrieve the cached value in the view.

[hancush]

Here's how to remove the total records text, via @beamalsky: https://github.com/datamade/committee-oversight/pull/137/files#diff-3965f1be15bd00793039b4e8259d444dR46-R48