Update Folder 4 for v24.3.0 of CIL

[lauramurgatroyd]

• Closes Check and update Folder 4 for v24.3.0 #227

Changes:

• finished incomplete sentence
• change objective_function to f
• added warnings filter to hide deprecation warnings

[hrobarts]

Hi Laura, changes look good to me.
In callbacks

• Just a note that while testing in a conda environment on reconVM1 these still ran slowly (7 minutes per example) which made me realise maybe we could run a lot fewer iterations for some of these examples without changing the learning outcome. I tried changing the iterations in this PR Switch to astra and reduce CPU use in deep dive demos #207.
• Typo

  CIL-Demos/demos/4_Deep_Dives/01_callbacks.ipynb

  Line 630 in 162a4ee

  "We seen that over the whole image, the mean squared error initially decreases, but then eventually starts to overfit to the noise in the data. "

  
  seen -> see

In preconditioner step sizes

• CIL-Demos/demos/4_Deep_Dives/04_preconditioner_stepsize.ipynb

  Line 201 in 162a4ee

  "Gradient descent with a fixed step size is not guaranteed to converge. As the iterations get near the minimum, a fixed step could cause iterations oscillate around the minimmum but not converge. \n",

  
  suggest update to "a fixed step could cause iterations to oscillate around the minimum but not converge"

[lauramurgatroyd]

Thanks Hannah, I will take a look at your PR and update the text as you suggested

[lauramurgatroyd]

Lets leave the #207 as a priority for the next online training

[hrobarts]

Thanks Laura, looks good to me