Are LDOs more efficient than switching regulators in very low currents?

Do the math.

At 5 V in and 1.2 V out, a linear regulator is limited to (1.2 V)/(5.0 V) = 24% efficiency. At 3 V in, the limit is 40%. The actual efficiency will be little less due to the current that the regulator itself uses to operate, but compared to 150 mA that will be negligible with pretty much any modern regulator.

So, the question becomes can a switcher do better than 40% efficiency in going from 5 V to 1.2 V? Absolutely. I'd expect to find switchers that can do 80% or better with 3 to 5 V in and 1.2 V out at 150 mA.

At these low voltages, you definitely need synchronous rectification to achieve good efficiency, but that will be built into any competent chip aimed at this kind of operating point. At 150 mA, you should be able to find a chip with the switching element built-in, which also helps with synchronous rectification. You will need to supply the inductor, input and output caps, and possibly a charge pump cap for making a higher voltage to drive the FET switch gate with. The datasheet will give you all the details. Read it carefully, then do what it says.

Multiple manufacturers should have products that do what you want. Some of the usual suspects are TI, Microchip, and Linear.

The efficiency of the LDO is a function of the input and output voltages.

$$\eta = \frac {P_{out}} {P_{in}} = \frac {I_{out} V_{out}} {I_{in} V_{in}}$$

$I_{out} \approx I_{in}$ for a series linear regulator ¹ ². Input and output currents cancel.

$$\eta \approx \frac {V_{out}} {V_{in}}$$

In your case where V_out is 1.2V, V_in is between 3V and 5V, the efficiency of the LDO will be between 40% and 24%. Any half-decent buck converter would have at least 80% efficiency at the expense of parts cost, switching noise, PCB real estate, engineering risk. An LDO with low efficiency can still be used when current is insignificant. Threshold of insignificance is decided individually for each end product.

¹ LDO's own quiescent current becomes significant when calculating standby current consumption. On the scale of 100mA current draw the LDO's quiescent current is insignificant.

² Series regulators, as opposed to the less common shunt regulators.

One common solution for some 3 LDO's is to use a Buck regulator followed by a very low dropout PFET linear regulator combined into one part. IC's with very low V logic-level FETs (Vt=0.65) are used in a 1.2V supply. Low noise is preferred for Vref and Va on ADCs used by some uC's.

The voltage ratio (%Vo/Vin) is a good estimate of the efficiency for LDO's except for some minimum load current required.

The choice comes down to your specs for; cost, complexity, efficiency and ripple. Low ripple SMPS demand the utmost care in following design guidelines from the OEM for BOM selection, cooling and power layout.