Add transformed coords of bombs to input feed

[Henrik Reinstädtler]

Resulting is a 5x5 matrix as input adding the transformed 💣, so we can train to avoid them

Plus reducing the number of displays of own position to one

[Tom Rix]

The 5x5 matrix is filled as follows with:

• arena field value for positions 4 steps in each direction (left, right, up, down) from current position (values as described in env code)
• arena field value for current position (values as described in env code)
• bomb locations (x and y values) of each of the four players (default value for no bomb is 0)

bomb1_x	up_1	up_2	up_3	up_4
bomb2_x	left_1	left_2	left_3	left_4
bomb3_x	right_1	right_2	right_3	right_4
bomb4_x	down_1	down_2	down_3	down_4
item_at_current_position	bomb1_y	bomb2_y	bomb3_y	bomb4_y