optimisation: pre-allocate, and fewer slices during summarisation

[rade]

Pre-allocating slices really pays dividends when we create loads of them and they don't grow very large.

Also, a bunch of code in templates/tables used slices as Maybes, i.e. returning nil or a single-element slice, which is horrendously inefficient.

[rade]

Benchmarks from running

go test -tags 'netgo unsafe' -run=x -cpu=2 -benchmem -bench='Summarize' --bench-report-path=/tmp/dev-on-dev-report.json

master

BenchmarkSummarizeHosts-2         	   10000	    158150 ns/op	   49524 B/op	     394 allocs/op
BenchmarkSummarizeControllers-2   	    2000	   1061920 ns/op	  348425 B/op	    3447 allocs/op
BenchmarkSummarizePods-2          	     500	   2471461 ns/op	  828257 B/op	    7548 allocs/op
BenchmarkSummarizeContainers-2    	     100	  30202501 ns/op	11220388 B/op	   84016 allocs/op
BenchmarkSummarizeProcesses-2     	     500	   2716148 ns/op	  881743 B/op	    7866 allocs/op

branch

BenchmarkSummarizeHosts-2         	   20000	     90850 ns/op	   25909 B/op	     205 allocs/op
BenchmarkSummarizeControllers-2   	    3000	    594095 ns/op	  199850 B/op	    2009 allocs/op
BenchmarkSummarizePods-2          	     500	   2076888 ns/op	  579624 B/op	    5468 allocs/op
BenchmarkSummarizeContainers-2    	     100	  27914949 ns/op	 9750272 B/op	   73029 allocs/op
BenchmarkSummarizeProcesses-2     	    1000	   2232948 ns/op	  556408 B/op	    4990 allocs/op

[rade]

I've merged this since it's all very straightforward.