I want to make a little night light with those tiny neon indicators.

I want to run 5 of them in 220V, but I don't know if I should use 1/4 watt resistors or more, and their value, I also want to know if I should run all them in series or in parallel with a single resistor.

1/4W resistors are fine, for 220-240V use 100㏀. I always use one resistor per neon though, never tried them together.

What type of the indicators?

There were "standard" and then "high brightness" types, they differ in current rating.
I wont drive them any more than really needed fortput you need, as the lifetime is inversely proportional to I^4 or even I^8 for the high output ones, so halving the current means 16x or 256x their rated life, so with few 100's uA it may become practically infinite.

But 1mA and below is pretty safe for all yet gives off pretty strong light, so about 150kOhm for a single lamp or 100kOhm for a series pair.
You won't be able to use more than 2 in series on 230V, that is already the edge (according to what I've tried).
It depends, if they are operated in complete darkness, or there is some light around, it makes quite significant difference in their ignition voltage.

How many years it will last if I use a 100k resistors for a pair? What resistor should I use for make it "infinite"?

https://datasheet.datasheetarchive.com/originals/scans/Scans-058/DSA2IH0031138.pdf
https://vcclite.com/wp-content/uploads/wpallimport/files/files/Neon%20Indicator%20Lamps%20v2.pdf
https://static1.squarespace.com/static/57ae25c75016e1690c41b053/t/57ae2fd003596ed24fa8f54c/1471033297306/Glow+Lamp+Specifications.pdf

Look for NE-2 for standard (0.6mA) and NE-2H/NE-2U for high brightness (1.9mA) type
When two in series at 230V, use column for 120V and double the resistance.
And from this point you may vary the resistance further according the brightness vs life compromise you need:

ExpectedLife = RatedLife * (RatedCurrent/OperatingCurrent) ^ 4 = RatedLife * (UsedResistor/RatedResistor)^4
ExpectedBrightness = RatedBrightness * (OperatingCurrent/RatedCurrent) = RatedBrightness * (RatedResistor/UsedResistor)


Edit: Found a white paper discussing the issue:
https://vcclite.com/wp-content/uploads/2017/02/Neon-Application-Information.pdf

You can use the bigger Neons from FL starters - At such low currents they will never heat up enough to close the contact. (And use resistors that can withstand the full 230V anyway to be on the safe side)

Can I light up a starter with a 200k resistor? I think 1mA will not heat the electrodes enough.

And use resistors that can withstand the full 230V anyway to be on the safe side

All will be sealed up in an incandescent bulb jacket, so, using normal 1/4 watt resistor should not me a problem. Unless there is a massive shaking and I accidentally put the wires close.

If you want just glow (without flashing), then the current gotta be low enough to not heat the bimetal too much. At 1mA the power dissipation will be around 100..150mW. If you use a starter that have only 1 bimetal electrode, you can use a diode in series with it to limit the discharge only to the non bimetal electrode, so to allow a bit higher current still without heating it too much

Can I light up a starter with a 200k resistor? I think 1mA will not heat the electrodes enough.
All will be sealed up in an incandescent bulb jacket, so, using normal 1/4 watt resistor should not me a problem. Unless there is a massive shaking and I accidentally put the wires close.

The problem is if you exceed the resistor voltage rating, it will flashover. The higher value resistors of a certain size are not anymore limited by the power, but by the voltage rating. The carbon "1/4W" 0207 size ones are usually limited to 250V peak, so not enough for 230V mains. The metal film 0.5W ones of the same 0207 are 500V, so just marginally OK. But you may connect multiple resistors in series to split the voltage...
And the power rating assumes 25degC ambient temperature (the air around the resistor), but a derating has to be applied as a strsight line dropping to zero at 70degC for carbon or 150degC for metal film.
And a good design practice dictates to never go above 50% of the power dissipation rating...