@oshpark boards are here!

What did I say earlier in the project? 

RTFD*...

*Read the Flipping Datasheet...

In hastily designing the PCB for the first post-breadboard version, I guessed which SOIC 8 footprint to use in KiCAD for the ATtiny85. And I lost that game. Even with bent-under pins, like others have done successfully, I wasn't sure I had soldered a connection on each pin.

Not only that, but by following blindly the idea to use two digital output pins to drive the MOSFET, I ended up connecting the two pins which play the role of  MISO and SCK during flashing of the code to the chip. So, my nifty SOIC8 clip couldn't program the ATtiny85 in-circuit.

So, I snipped the legs and desoldered the ATtin85 (there was no way of saving it). I lost a non-critical pad from the PCB footprint in the process. This time I programmed a new ATtiny85 before attempting to solder its subsequently bent legs to the board. Progress was made in burning an 8MHz bootloader and then the Arduino sketch to the chip. Solder. Apply power. Nothing happened...

So, I tried driving the breadboard version with the same code running on a DIP ATtiny85 (not shown in the reminder picture below). 

It did absolutely nothing with the LEDs. At this point, I have code that's not working, solder connections I don't have confidence in and a circuit that I'm not sure I have routed properly, even if you ignore the connection of MOSI and SCK nets. Too many variables!

So, I worked on the code for the DIP ATtiny85, as I know that circuit works. Sorted. But the multiple (I tried a second time) attempts to attach an SMD ATtiny85 meant that the pads were pretty much ruined (and small for green-wiring to)  so I green wired a signal to the MOSFET from the DIP ATtiny85 and applied VCC and Ground to the header pin holes I had left for this purpose (although not correctly spaced) - SUCCESS!The flash is between ON and ALWAYS ON, which has a slight glow to achieve the aim of being easily found in the dark.

I also got some PTC resettable fuses based the mulling I did in a previous log (7.4 Ohm initial resistance). Although I wasn't expecting a noticeable difference in brightness because of the SPICE simulation, I'm afraid to say I did notice one. I wired a clipped piece of resistor leg in the holes to make the circuit in the GIF you see above.

Now, as I revise the PCB, I am working to the constraints of the small (25x25x50mm) aluminium enclosures I'd like to use.

The TP4056 board fits nicely but the rotated LEDs on V0.1 PCB above aren't going to fit, so I'm likely to bunch them up into a more conventional recti-linear array. I'll also switch the footprint to use an SMD fuse and allow more freedom on the reverse of the PCB. The other major consideration in the next version will be the battery selection - likely to lead to a charge rate resistor change on the TP4056 module to ensure we're still charging at around 1C. This will leave the button, charging terminals and ingress protection for another day. The magnet will also need inserting to get it working when I get the charging station going.

Also, for this revision I'm going to try the OpenGL mode in KiCAD - thanks for the tip, @Chris Gammell on a recent episode of the Amp Hour!