Clinic roster merge

Merge two exports of a clinic's patient roster into one clean CSV.

Each file is a list of lines; the first line is a header. Cells are separated by commas. A cell may be wrapped in double quotes, in which case it can contain commas and a doubled "" stands for one " character. A quote that does not open a cell is an ordinary character. Strip leading and trailing spaces from every cell value after unquoting.

The four fields, in output order, are id, name, phone and site. Identify a header cell by deleting its spaces, underscores and hyphens and lowercasing it:

• id: id, patientid, patientnumber, mrn
• name: name, fullname, patientname
• phone: phone, tel, telephone, mobile
• site: site, clinic, location

Other columns are ignored. A file with no id column contributes nothing. A field a file lacks, or a cell missing because its line is short, is empty; extra cells beyond the header are ignored.

Phone numbers keep digits only. Patients are identified by id (compared exactly as text, so 007 and 7 differ); lines whose id is empty are dropped.

Process all lines of csv_a, then all lines of csv_b, in order. Each line is laid over the patient's record: every non-empty cell replaces what the record has for that field, and empty cells change nothing.

Input: csv_a and csv_b, lists of strings.

Output: a list of CSV lines: the header id,name,phone,site, then one line per patient, sorted by id in ascending character order. Write a cell inside double quotes (doubling any " it contains) only if it contains a comma or a double quote; always write all four cells.