iPhones/iPod Touches/iPads all have a Unified Memory Architecture which mean that both the CPU and GPU share system memory. There is no dedicated video memory on these devices.
The advantage is that you don't need to worry about running out of video memory for your textures or vertex data (your app will be terminated by iOS for using too much memory before that happens). The disadvantage is that you share the same memory bandwidth for gameplay and graphics. The more memory bandwidth you dedicate to graphics, the less you will have for gameplay and physics
Your information is very good for RAM and CPU, but I don't see anything amiss and things look quite balanced. Yes, there is one program that's busy and not responding, so you could kill that to rule it out as something that's slowing down Photoshop.
My hunch is you will need to examine storage to find the slow down.
My reasoning is Photoshop does a good job of pre-allocating disk space for swap of it's own (so it doesn't even swap to the OS swap) as well as pre-allocating all the RAM it thinks it will need. Even without seeing the normal RAM and CPU charts you posted, it's unlikely your Mac is paging virtual memory to disk based on the stats you posted from a snapshot in time.
Open terminal app from /Applications/Utilities and run two commands to list the attached drives (so you know which is disk0, disk1 etc...) and another to monitor the actual input and output for the drive.
df | grep disk
iostat 30
The second command will run continuously until you press control+c or close the terminal app or window.
My guess is you will see that there is continuous load on one or more disks and quitting the programs that are causing the disk to be busy will make Photoshop more snappy. You've already ruled out high CPU load and memory contention (the pressure would be yellow or red if that were to slow down Photoshop) so the next likely candidate is that the system is waiting for the storage.
The other suggestions to try killing other applications are good ones, but you will be more successful if you can measure the exact slow item and then change things and re-measure later.
Here is a sample of what you might see on an idle system:
Mac:~ me$ iostat 30
disk0 disk1 cpu load average
KB/t tps MB/s KB/t tps MB/s us sy id 1m 5m 15m
36.80 20 0.71 24.99 0 0.00 6 3 91 1.34 1.61 1.65
32.00 0 0.01 0.00 0 0.00 1 1 97 1.32 1.60 1.65
0.00 0 0.00 0.00 0 0.00 1 2 97 1.37 1.61 1.65
4.00 0 0.00 0.00 0 0.00 1 2 97 1.31 1.58 1.64
28.57 1 0.04 0.00 0 0.00 4 3 93 1.48 1.61 1.65
9.85 3 0.02 0.00 0 0.00 2 1 97 1.44 1.60 1.65
Here is what first a minute of activity on the internal SSD then idle, then a minute of activity on the Time Machine drive and idle, and finally Time Machine running with both drives having equivalent activity (read from one = write to another):
Mac:~ me$ iostat 30
disk0 disk1 cpu load average
KB/t tps MB/s KB/t tps MB/s us sy id 1m 5m 15m
36.95 20 0.71 42.93 0 0.00 6 3 91 1.99 1.90 1.75
8.1 1266 10.09 8.00 0 0.00 8 12 81 1.94 1.89 1.75
10.18 774 7.69 0.00 0 0.00 22 9 69 1.64 1.82 1.73
18.42 125 2.25 0.00 0 0.00 11 4 86 1.50 1.77 1.71
32.27 21 0.67 0.00 0 0.00 6 2 92 1.40 1.73 1.70
12.29 48 0.57 6.78 223 1.47 23 8 70 1.74 1.78 1.72
20.30 14 0.28 9.61 53 0.50 2 3 95 1.39 1.69 1.69
17.73 5 0.08 14.75 0 0.01 11 2 87 1.23 1.62 1.66
32.30 23 0.73 0.00 0 0.00 4 2 94 1.50 1.67 1.68
13.14 3 0.04 0.00 0 0.00 6 1 92 1.45 1.64 1.66
21.07 24 0.50 6.12 0 0.00 9 3 88 2.23 1.80 1.72
40.77 197 7.86 37.93 202 7.47 13 10 77 2.27 1.83 1.73
Best Answer
There has often been problems with AMD based GPUs and their reporting of video memory usage in iStat Menu. This has in the past been for example always reporting 100% usage, always reporting a climbing usage, etc.
If you haven't got any performance degradation, and your computer usage really isn't affected by this - I would just ignore the readout.
There's no tool to do a manual release - it doesn't really make sense. If the VRAM is actually in use, "manually releasing" it would make programs crash or behave unpredictable - nobody wants that.
If you really want you could probably lower the VRAM usage by logging out, or by killing WindowServer in its entirety. There's just no real reason to do so - unless you're experiencing a performance degradation, and then it would be better to analyze that specifically.