Different URLs that contains “www” in the host name are incorrectly canonicalized.

The following URLs are incorrectly canonicalized with SURT as "com)/".
```
SURT = "com)/"
1. https://www1355544.com/
2. https://www3288.com/
3. https://www504778.com/
4. https://www556798.com/
5. https://www57912.com/
```

Problem is that 'www\d*\.' is removed from the URL, which leaves "com)/" as the SURT.
https://github.com/internetarchive/surt/blob/6934c321b3e2f66af9c001d882475949f00570c5/surt/IAURLCanonicalizer.py#L127

the other instance where the URLs were correctly canonicalized by removing 'www\d*\.'
```
SURT = "jp,co,daily)/"
1. https://daily.co.jp/
2. https://www4.daily.co.jp/
3. https://www8.daily.co.jp/
4. https://www9.daily.co.jp/
```
We proposed to update the regex where it should only remove the 'www\d*\.' if at least two dots are present in the hostname.

Start in surt/IAURLCanonicalizer.py around line 127, where the www-number hostname prefix is removed. Check the regex against the listed single-label and multi-label hostnames, then verify that only the intended prefixes are removed and the resulting SURT values are correct for all examples.

Written by the indexing model from the issue text.