[Fix] JHUData.records(): complement is done after subsetting with start date

[lisphilar]

Summary

Because complement is done after subsetting with start date, un-expected date is sometimes selected as the first date of JHUData.records(). The output of complement is dependent of the start date when subsetting.

For example, "01May2020" is expected as the first date with jhu_data.records(country="Japan", start_date="01May2020"), but "02May2020" is returned.

Codes and outputs

https://gist.github.com/lisphilar/fcecfa840a1f14acdb706508cddda6fa

Environment

• CovsirPhy version: 2.22.0-alpha
• Python version: 3.7.11
• Installation: pip
• System: Google Colaboratory

[lisphilar]

With #922, "01May2020" will be the first date.

jhu_data.records(country="Japan", start_date="01May2020")[0].head(20)

	Date	Confirmed	Infected	Fatal	Recovered	Susceptible
0	2020/5/1	1454	305	402	747	1.27E+08
1	2020/5/2	1481	259	426	796	1.27E+08
2	2020/5/3	1511	199	458	854	1.27E+08
3	2020/5/4	1533	171	475	887	1.27E+08
4	2020/5/5	1551	178	485	888	1.27E+08
5	2020/5/6	1564	170	505	889	1.27E+08
6	2020/5/7	1575	172	513	890	1.27E+08
7	2020/5/8	1583	173	518	892	1.27E+08
8	2020/5/9	1594	143	558	893	1.27E+08
9	2020/5/10	1604	141	570	893	1.27E+08
10	2020/5/11	1609	138	578	893	1.27E+08
11	2020/5/12	1616	124	598	894	1.27E+08
12	2020/5/13	1632	117	621	894	1.27E+08
13	2020/5/14	1637	104	639	894	1.27E+08
14	2020/5/15	1649	92	660	897	1.27E+08
15	2020/5/16	1653	81	674	898	1.27E+08
16	2020/5/17	1658	66	692	900	1.27E+08
17	2020/5/18	1660	61	697	902	1.27E+08
18	2020/5/19	1666	51	710	905	1.27E+08
19	2020/5/20	1668	45	717	906	1.27E+08