Why flag is of 8-bit in the case of bit and byte oriented protocol ?

it depends on your application

Sorry William i doesn't depend on application,
The the 8 bit flag is set to tackle problem of variable size framing. As we know a frame in data link layer may consists of header+trailer along with information about payload, error correction, detection and redundant bits. And they all are in multiple of 8 bits. So it might happen that payload may get mis-undertaken as any other data if matching bits are found so to separate each frame 8 extra bits are flagged on either sides of frame to differentiate it from other frames. A flag consists of start or end of frame and some other details. I hope that was easy explanation.
Guys specially from Networking field please correct if something went wrong.

Yes.. Not only flags, but most of the fields in many protocols, like payload length, time to live etc.. are made as 8 bits or a multiple of 8 bits, so that the total no. of bits will be a multiple of 8.
In some protocols, if the payload length is not a multiple of 8, then zeroes are padded to the end to make it a multiple of 8, for the same reason.