Performance optimization by deferred deserialization

pg_roaringbitmap v0.5.2 optimized performance of some functions through deferred deserialization, that is, use new roaring_buffer_t instead of roaring_bitmap_t.

Refer to below for the effect of optimization

benchmark script

benchmark/benchmark.sql

v0.5.1 vs v0.5.2(PG12.3)

No	test	v0.5.1(ms)	v0.5.2(ms)
1	rb_and_1	186.808	158.392
2	rb_and_2	205.653	64.741
3	rb_or_1	217.008	101.657
4	rb_or_2	155.667	65.872
5	rb_xor_1	266.933	111.47
6	rb_xor_2	437.087	236.536
7	rb_andnot_1	152.366	85.031
8	rb_andnot_2	177.482	50.528
9	rb_add	109.252	67.742
10	rb_contains_1	156.08	5.248
11	rb_contains_2	177.243	28.861
12	rb_contains_3	127.605	21.818
13	rb_intersect_1	10.558	3.474
14	rb_intersect_2	60.136	18.502
15	rb_equals_1	11.616	3.445
16	rb_equals_2	81.01	17.533
17	rb_andnot_1	163.378	88.159
18	rb_andnot_2	102.951	53.082
19	rb_index_1	93.413	18.91
20	rb_cardinality_1	89.545	27.521
21	rb_and_cardinality_1	151.704	86.746
22	rb_and_cardinality_2	96.901	33.94
23	rb_or_cardinality_1	107.328	83.086
24	rb_or_cardinality_2	43.405	48.706
25	rb_xor_cardinality_1	95.664	79.639
26	rb_xor_cardinality_2	83.32	38.438
27	rb_andnot_cardinality_1	89.864	78.376
28	rb_andnot_cardinality_2	96.584	54.209
29	rb_is_empty_1	75.312	22.18
30	rb_range_1	28.516	38.503
31	rb_range_cardinality_1	76.501	110.586
32	rb_min_1	38.235	20.497
33	rb_max_1	57.786	17.423
34	rb_rank_1	52.568	27.182
35	rb_jaccard_dist_1	111.118	74.05
36	rb_jaccard_dist_2	60.378	26.698
37	rb_to_array_1	88.361	140.412
38	rb_iterate_1	78.836	96.115
39	rb_build_agg_1	236.013	267.927
40	rb_or_agg_1	29.846	32.957
41	rb_and_agg_1	1.379	1.262
42	rb_xor_agg_1	28.762	34.628
43	rb_or_cardinality_agg_1	27.907	33.042
44	rb_and_cardinality_agg_1	1.341	1.37
45	rb_xor_cardinality_agg_1	33.459	31.767