[DaudAhmadJF]

Add --output csv to every command, alongside pretty, raw, table and yaml.

A top-level JSON array becomes one row per record, which is what the bulk endpoints return; any other response becomes a single row. Nested objects flatten into dotted column names such as location.city. An array of scalars joins into one cell with |, while an array of objects keeps a column per element so its fields stay addressable. Columns are the sorted union of the keys of every row, so the same response always produces the same file.

Numbers are read with json.Decoder.UseNumber and written exactly as the API sent them. Decoding into float64 would render current_time_unix as 1.772535675329e+09 and round long identifiers.

Values are guarded against spreadsheet formula injection, since organization names and whois text come from third parties. A leading =, @, tab or carriage return is quoted off, as is a + or - followed by a non-digit. Signed data is left intact, so the -122.08421 longitude, the +1 calling code, the +1.00H and -1.00H DST durations and the lone - the astronomy endpoint uses for an absent sun or moon status all survive unchanged.

On the bulk commands --output-file now follows --output: csv saves a .csv file, every other format keeps saving the .json it always has. The extension is appended only when the given name lacks it, so --output-file results and --output-file results.csv both write results.csv. This also fixes --output-file=output.json, which previously wrote output.json.json.