The problem is that your analysis is all done from the perspective of the frame that measures the box to be moving at 0.1c -- in this frame, it's true that the time for the light to get from the source to the wall is different from the time for the light to get from the wall back to the source. But if these same events are measured by someone inside the box (or just moving alongside it and looking in), using rulers and clocks at rest relative to the box (and with clocks at either end of the box synchronized in the box's rest frame using the Einstein synchronization convention), then the person inside the box measures the same time for the light to get from the source to the wall as he measures for the light to get from the wall back to the source. To see this you must take into account three different ways in which the rulers and clocks of the person inside the box seem "off" from the perspective of the person who sees the box moving at 0.1c: length contraction, time dilation and the relativity of simultaneity.
Let's call the person who sees the box moving at 0.1c "Alice", and the person inside the box (and at rest relative to it) "Bob". Then length contraction means that if the box has a length of 300,000 km in its own rest frame as measured by Bob's rulers, Alice sees the length of the box as shorter by a factor of $ \sqrt{1 - v^2/c^2} $, in this case $ \sqrt{1 - (0.1c/c)^2} $ = 0.9949874 (and she also see's Bob's rulers as shrunken), so she measures the box to have a length of only 0.9949874 * 300,000 km = 298,496.2 km. Time dilation means that Alice would see Bob's clocks ticking more slowly than hers, so the length of each tick of Bob's clock is longer by a factor of $ 1 / \sqrt{1 - v^2/c^2}$, which is equivalent to saying that if t seconds go by on Alice's clock, Alice measures only $ t * \sqrt{1 - v^2/c^2} $ seconds to have gone by on Bob's clock, or 0.9949874 seconds on Bob's clock for every second that goes by on Alice's clock, as measured in Alice's frame. And the relativity of simultaneity means that if Bob had "synchronized" clocks at either end of the box in his frame, and measured them to be at a distance of 300,000 km apart or 1 light-seconds apart (I will assume as you did that the speed of light is exactly 300,000 km/second even though this is not quite right), then in Alice's frame the two clocks are slightly out-of-sync, at any given moment the clock at the back end of the box (where the source is located) shows a reading that's ahead of the clock at the front of the box (where the reflecting wall is located) by (0.1c)(1 light-seconds)/c^2 = 0.1 seconds.
Say in Alice's frame the back end of the box is at position x=0 at t=0 in terms of her space and time coordinates, and at that moment it emits light. Since the box has a length of 298,496.2 km in her frame, the front of the box must be at x=298,496.2 km at t=0. The position of the front of the box as a function of time in Alice's frame is 30,000*t + 298,496.2, and the position of the light emitted from the source as a function of time is 300,000*t, so the light will catch up with the source when 300,000*t = 30,000*t + 298,496.2, or t=1.105541 seconds, at position x=331662.3 km in Alice's frame. Then since the position of the back of the box as a function of time is 30,000*t, and the position of the reflected light as a function of time is -300,000*(t - 1.105541) + 331662.3, the reflected light will get back to the source when 30,000*t = -300,000*(t - 1.105541) + 331662.3, or when t=2.010075 seconds. Thus in Alice's frame the light took 1.105541 seconds to get from the source at the back of the box to the wall at the front, and it took (2.011075 - 1.105541) = 0.905534 seconds to get back to the source after being reflected by the front wall.
So in Alice's frame, at the moment the light is emitted by the source, let's say Bob's clock next to the source reads T=0 seconds, in which case Bob's clock at the other end of the box will be behind by 0.1 seconds (due to the relativity of simultaneity discussed above), reading T=-0.1 seconds. Then if Alice measures a time of 1.105541 seconds for the light to reach the front end of the box, each of Bob's two clocks will only have ticked forward by 0.9949874 * 1.105541 seconds = 1.1 seconds (due to time dilation). So in Alice's frame, at the moment the light reflects off the front wall, the clock at the back of the box next to the source reads T = 0 + 1.1 = 1.1 seconds, while the clock at the front wall where the light is striking reads T = -0.1 + 1.1 = 1.0 seconds. So, Bob measures the ship to be 300,000 km long, and he measures the light to have departed the source at T=0 seconds and to have arrived at the front wall at T=1.0 seconds, meaning he measures the light to have traveled at exactly 300,000 km/s in moving from theback wall (where the source is) to the front wall. Likewise, if the light is reflected and returns to the back wall where the source is located at t=2.011075 seconds in Alice's frame, during that time Alice sees Bob's clock to have only ticked forward by 0.9949874 * 2.011075 = 2.0 seconds, so the light returns to the source when Bob's clock there shows a time of 2.0 seconds. So, since Bob measures the ship to be 300,000 km long, and he measures the light to have been reflected from the front wall at T=1.0 seconds and to have returned to the source on the back wall at T=2.0 seconds, that must mean he also measured the light to have taken 1 second to travel 300,000 km when moving from the front wall to the back wall.
Finally, note that we could make the situation completely symmetrical by imagining a similar box at rest in Alice's frame, with Alice measuring it to be 300,000 km long by her rulers, and with her having two clocks at the front and back which were at rest in her frame and synchronized according to the Einstein synchronization convention. Then Bob would similarly measure Alice's rulers to be shrunken, and her clocks running slow and out-of-sync, and if she sent a light beam from one end of the box to the other and back, she would measure the light ray to take 1 second in each direction using her clocks while Bob would measure it to take 1.105541 seconds in one direction and 0.905534 seconds in the other, using his own clocks.