Collapse numeric flag types to be like strconv

[meatballhat]

The strconv API with Parse.* functions omits the bit size and always returns the 64-bit type. Doing the same for the relevant flag types would reduce the library surface area and testing code significantly (imho)

• collapse Int64Flag into IntFlag with int64 destination/return type #1763
• collapse Int64SliceFlag into IntSliceFlag with []int64 destination/return type #1764
• collapse Uint64Flag into UintFlag with uint64 destination/return type #1766
• collapse Uint64SliceFlag into UintSliceFlag with []uint64 destination/return type #1767
• rename Float64Flag to FloatFlag retaining float64 destination/return type #1771
• rename Float64SliceFlag to FloatSliceFlag retaining []float64 destination/return type #1772