modified freelist_hmap/hashmapGetFreePageIDs function with better performance

[nicktming]

Since forwardMap key is the start point, value is the length starting with the key, then we just sort the key to make the whole array sorted instead of sorting the whole array.

This is my performance test compared with these two algorithms:

1. N: is the number of forwardMap key
2. val: is the value of each key
3. time_used_origin: is the time used of the origin function
4. time_used_new: is the time used of the improved function

N	val	time_used_origin	time_used_new
10000	1	2.301202ms	2.614463ms
10000	10	18.204936ms	2.942327ms
10000	100	164.167666ms	7.275755ms
10000	1000	1.71088519s	115.790789ms
10000	10000	19.85861009s	1.625219654s

func main() {
	N := int32(10000)
	fm := make(map[pgid]uint64)
	i := int32(0)
	val := int32(0)
	for i = 0; i < N;  {
		val = rand.Int31n(1000)
		fm[pgid(i)] = uint64(val)
		i += val
	}

	f := freelist{
		forwardMap: fm,
	}
	start := time.Now()
	f.hashmapGetFreePageIDs()
	end := time.Now()
	fmt.Printf("origin time:%v\n", end.Sub(start))

	start = time.Now()
	res := f.newHashmapGetFreePageIDs()
	end = time.Now()
	fmt.Printf("new time:%v\n", end.Sub(start))


	if !sort.SliceIsSorted(res, func(i, j int) bool { return res[i] < res[j] }) {
		panic("pgids not sorted")
	}
}

based on the above testing program, when changing N, got the following result compared with two functions

ENV : 8 GB 2133 MHz LPDDR3 / 1.6 GHz 2 Cores Intel Core i5

N	time_used_origin	time_used_new
100	71.608µs	9.022µs
1000	226.745µs	19.359µs
10000	1.290836ms	77.194µs
100000	15.973611ms	777.995µs
1000000	160.912048ms	5.714576ms
10000000	1.789962277s	99.443407ms

[ahrtr]

Suggested change

	keys := make([]pgid, 0, len(f.forwardMap))
	startPageIds := make([]pgid, 0, len(f.forwardMap))

[ahrtr]

Suggested change

	size, ok := f.forwardMap[start]
	if ok {
	if size, ok := f.forwardMap[start]; ok {

[ahrtr]

This PR looks good to me. Please

1. squash the commit and signoff the final commit;
2. rebase this PR (Note: we added some github workflows).

cc @ptabor @xiang90

[ahrtr]

Added label resurrection because there is no response from the original author. Please anyone else feel free to resurrect this PR.

[cenkalti]

Rebased the commits and sent as new PR #419

[ahrtr]

superseded by #419