Fix (examples): adding bias_quant to final linear layer in resnet18

[i-colbert]

Need to quantize the bias of the final layer to support streamlining in end-to-end FINN flow. Verified that this change does not impact accuracy on CIFAR10.

...
2023-10-11 16:06:12,585 Test: [97/100]  Model Time 0.109 (0.121)        Loss Time 0.000 (0.000) Loss 2.6345 (2.2591)    Prec@1 92.000 (92.592)  Prec@5 99.000 (99.694)
2023-10-11 16:06:12,690 Test: [98/100]  Model Time 0.094 (0.120)        Loss Time 0.011 (0.000) Loss 1.7418 (2.2538)    Prec@1 94.000 (92.606)  Prec@5 100.000 (99.697)
2023-10-11 16:06:12,790 Test: [99/100]  Model Time 0.100 (0.120)        Loss Time 0.000 (0.000) Loss 3.5667 (2.2670)    Prec@1 92.000 (92.600)  Prec@5 100.000 (99.700)