covid19-sir: [Fix] too long delay period of Hungary with Scenario.estimate_delay()

Summary

As discussed in #670, Scenario.estimate_delay() returns 53 days as delay period of hungary and this is too long. This causes change points detection failure in S-R trend analysis.

We may need to revise Scenario.estimate_delay() to get many data for “Period Length” calculation. We have only 11 records for Hungary as an example.

Codes

snl = cs.Scenario(country="Hungary", province=None)
snl.register(jhu_data, population_data, extras=[oxcgrt_data])
# 53, dataframe
delay, df = snl.estimate_delay()
# S-R trend analysis
_ = snl.trend()

Outputs

Index	Confirmed	Stringency_index	Period Length
0	48757	40.74	NaN
1	32	50	12
2	32	50	36
3	32	50	29
4	4183	54.63	17
5	94916	57.41	53
6	3990	61.11	31
7	3741	66.67	53
8	300	71.3	2
9	366279	72.22	5
10	2443	76.85	118
11	614612	79.63	19

Environment

• CovsirPhy version: 2.17.0-pi
• Python version: 3.9.2
• Installation: poetry
• System: WSL (Ubuntu)