Bandwidth.
A scope may have ~ 20 pF or input capacitance in parallel with 1 MΩ -- giving a time constant of about 20p*1M = 20 us. The photodiode (which has additional capacitance, further reducing the BW) generates a current and this takes some time to charge the 20 pF capacitance. When the light turns off, it takes some time for the 1 MΩ to discharge the capacitance. The final voltage is the photocurrent * 1 MΩ, or the VF of the photodiode itself (about 0.6 V). An opamp appears to short the photodiode current to the virtual ground -- thus that current doesn't have to charge any capacitance -- the opamp does the work of driving the output and also the virtual ground through the feedback resistor, giving an output voltage of Iphotodiode* Rfeedback.
You can see this yourself -- touch and remove a scope probe to a battery or power supply -- you will see a fast rising edge when you make contact, but an exponential decay when removed (it may be quite noisy when you do this manually, but you will be able to get a few clean traces)